OpenAI \u201cPentagon deal\u201d \u2014 timeline, red lines, and backlash signals<\/title>\r\n\r\n <style>\r\n \/* ===========================\r\n DAILLAC Article (Standalone)\r\n =========================== *\/\r\n\r\n :root{\r\n --dlx-bg:#ffffff;\r\n --dlx-text:#0f172a;\r\n --dlx-muted:rgba(15,23,42,.72);\r\n --dlx-border:rgba(16,24,40,.10);\r\n\r\n --dlx-surface:#f8fafc;\r\n --dlx-surface-2:#eef2ff;\r\n\r\n --dlx-accent:#0f766e;\r\n --dlx-accent-2:#8b5cf6;\r\n --dlx-accent-soft:rgba(139,92,246,.12);\r\n\r\n --dlx-radius:16px;\r\n --dlx-radius-sm:12px;\r\n\r\n --dlx-shadow: 0 10px 24px rgba(16,24,40,.06);\r\n --dlx-shadow-sm: 0 6px 14px rgba(16,24,40,.06);\r\n }\r\n\r\n *{ box-sizing:border-box; }\r\n html,body{ margin:0; padding:0; background:var(--dlx-bg); color:var(--dlx-text); }\r\n body{ font-family:system-ui,-apple-system,Segoe UI,Roboto,Arial,sans-serif; line-height:1.65; }\r\n\r\n a{ color:var(--dlx-accent); text-decoration:none; }\r\n a:hover{ text-decoration:underline; }\r\n\r\n \/* Accessibility helper *\/\r\n .dlx-sr-only{\r\n position:absolute !important;\r\n width:1px;height:1px;\r\n padding:0;margin:-1px;\r\n overflow:hidden;clip:rect(0,0,0,0);\r\n white-space:nowrap;border:0;\r\n }\r\n\r\n \/* Article container *\/\r\n .dlx-article{\r\n max-width: 980px;\r\n margin: 0 auto;\r\n padding: 22px 16px 56px;\r\n background: var(--dlx-bg);\r\n }\r\n\r\n \/* Hero *\/\r\n .dlx-article__hero{\r\n position:relative;\r\n padding: 18px 16px 18px;\r\n border: 1px solid var(--dlx-border);\r\n border-radius: var(--dlx-radius);\r\n background:\r\n radial-gradient(1200px 500px at 0% 0%, rgba(15,118,110,.10), transparent 55%),\r\n radial-gradient(900px 420px at 100% 0%, rgba(139,92,246,.10), transparent 60%),\r\n #fff;\r\n box-shadow: var(--dlx-shadow);\r\n }\r\n\r\n .dlx-article__eyebrow{\r\n display:inline-flex;\r\n gap:8px;\r\n align-items:center;\r\n margin: 0 0 10px;\r\n padding: 6px 10px;\r\n border-radius: 999px;\r\n border: 1px solid rgba(15,118,110,.22);\r\n background: rgba(15,118,110,.06);\r\n color: rgba(15,118,110,.95);\r\n font-weight: 700;\r\n font-size: 12px;\r\n letter-spacing: .02em;\r\n }\r\n\r\n .dlx-article h1{\r\n margin: 0 0 10px;\r\n font-size: 34px;\r\n line-height: 1.15;\r\n letter-spacing: -0.02em;\r\n }\r\n\r\n .dlx-article__lead{\r\n margin: 0 0 14px;\r\n color: var(--dlx-muted);\r\n font-size: 16px;\r\n max-width: 72ch;\r\n }\r\n\r\n .dlx-meta{\r\n display:flex;\r\n flex-wrap:wrap;\r\n gap:10px 14px;\r\n padding-top: 10px;\r\n border-top: 1px dashed rgba(15,23,42,.18);\r\n color: rgba(15,23,42,.72);\r\n font-size: 13px;\r\n }\r\n\r\n \/* TOC *\/\r\n .dlx-toc{\r\n margin: 18px 0 0;\r\n padding: 14px 14px;\r\n border: 1px solid var(--dlx-border);\r\n border-radius: var(--dlx-radius);\r\n background:\r\n linear-gradient(90deg, rgba(139,92,246,.08), rgba(15,118,110,.06)),\r\n #fff;\r\n box-shadow: var(--dlx-shadow-sm);\r\n }\r\n .dlx-toc__title{\r\n font-weight: 800;\r\n margin: 0 0 8px;\r\n color: rgba(15,23,42,.92);\r\n }\r\n .dlx-toc ul{\r\n margin:0;\r\n padding-left: 18px;\r\n }\r\n .dlx-toc li{ margin: 6px 0; }\r\n\r\n \/* Sections *\/\r\n .dlx-section{ margin-top: 22px; }\r\n .dlx-section h2{\r\n margin: 0 0 10px;\r\n font-size: 22px;\r\n letter-spacing: -0.01em;\r\n }\r\n .dlx-section h3{\r\n margin: 16px 0 8px;\r\n font-size: 16px;\r\n }\r\n\r\n .dlx-muted{ color: var(--dlx-muted); }\r\n\r\n \/* Layout helpers *\/\r\n .dlx-grid{ display:grid; gap: 12px; }\r\n .dlx-grid--2{ grid-template-columns: repeat(2, minmax(0,1fr)); }\r\n .dlx-grid--4{ grid-template-columns: repeat(4, minmax(0,1fr)); }\r\n @media (max-width: 860px){ .dlx-grid--4{ grid-template-columns: repeat(2, minmax(0,1fr)); } }\r\n @media (max-width: 640px){ .dlx-grid--2, .dlx-grid--4{ grid-template-columns: 1fr; } }\r\n\r\n \/* Cards \/ callouts *\/\r\n .dlx-card{\r\n border: 1px solid var(--dlx-border);\r\n border-radius: var(--dlx-radius);\r\n background: #fff;\r\n padding: 14px 14px;\r\n box-shadow: var(--dlx-shadow-sm);\r\n }\r\n\r\n .dlx-callout{\r\n border: 1px solid rgba(139,92,246,.22);\r\n border-radius: var(--dlx-radius);\r\n background:\r\n linear-gradient(90deg, rgba(139,92,246,.10), rgba(15,118,110,.06)),\r\n #fff;\r\n padding: 14px 14px;\r\n box-shadow: var(--dlx-shadow-sm);\r\n }\r\n .dlx-callout__title{\r\n font-weight: 900;\r\n margin: 0 0 6px;\r\n color: rgba(15,23,42,.92);\r\n }\r\n\r\n .dlx-note{\r\n border: 1px solid rgba(15,118,110,.20);\r\n border-radius: var(--dlx-radius);\r\n background:\r\n radial-gradient(800px 240px at 0% 0%, rgba(15,118,110,.10), transparent 60%),\r\n #fff;\r\n padding: 14px 14px;\r\n box-shadow: var(--dlx-shadow-sm);\r\n }\r\n .dlx-note__title{\r\n font-weight: 900;\r\n margin: 0 0 6px;\r\n color: rgba(15,23,42,.92);\r\n }\r\n\r\n .dlx-mb-0{ margin-bottom: 0 !important; }\r\n\r\n \/* KPI *\/\r\n .dlx-kpi{\r\n border: 1px solid var(--dlx-border);\r\n border-radius: var(--dlx-radius);\r\n background: #fff;\r\n padding: 12px 12px;\r\n box-shadow: var(--dlx-shadow-sm);\r\n }\r\n .dlx-kpi__value{\r\n display:block;\r\n font-weight: 950;\r\n font-size: 28px;\r\n line-height: 1.05;\r\n letter-spacing: -0.02em;\r\n background: linear-gradient(90deg, var(--dlx-accent), var(--dlx-accent-2));\r\n -webkit-background-clip: text;\r\n background-clip: text;\r\n color: transparent;\r\n margin-bottom: 6px;\r\n }\r\n .dlx-kpi__label{\r\n display:block;\r\n color: var(--dlx-muted);\r\n font-size: 12px;\r\n font-weight: 700;\r\n }\r\n\r\n \/* =======================\r\n TABLE (Timeline Improved)\r\n ======================= *\/\r\n .dlx-table-wrap--timeline{\r\n border:1px solid var(--dlx-border);\r\n border-radius: var(--dlx-radius);\r\n background:#fff;\r\n overflow:auto;\r\n -webkit-overflow-scrolling:touch;\r\n box-shadow: var(--dlx-shadow);\r\n }\r\n\r\n .dlx-table{\r\n width:100%;\r\n border-collapse:separate;\r\n border-spacing:0;\r\n table-layout:fixed;\r\n min-width: 920px;\r\n }\r\n\r\n .dlx-table thead th{\r\n position:sticky;\r\n top:0;\r\n z-index:2;\r\n text-align:left;\r\n padding:12px 12px;\r\n background:\r\n linear-gradient(90deg,\r\n rgba(15,118,110,.10),\r\n rgba(139,92,246,.08)\r\n ),\r\n #ffffff;\r\n border-bottom:1px solid var(--dlx-border);\r\n color:#0f172a;\r\n font-size: 13px;\r\n }\r\n\r\n .dlx-table tbody td{\r\n vertical-align:top;\r\n padding:12px 12px;\r\n border-bottom:1px solid rgba(16,24,40,.08);\r\n color:#0f172a;\r\n font-size: 13px;\r\n overflow-wrap:anywhere;\r\n word-break:break-word;\r\n }\r\n\r\n .dlx-table tbody tr:nth-child(odd) td{\r\n background:rgba(15,118,110,.04);\r\n }\r\n\r\n .dlx-table tbody tr:hover td{\r\n background:rgba(139,92,246,.08);\r\n }\r\n\r\n \/* Mobile: rows become cards *\/\r\n @media (max-width: 720px){\r\n .dlx-table{ min-width: 0; table-layout:auto; }\r\n .dlx-table thead{ display:none; }\r\n\r\n .dlx-table,\r\n .dlx-table tbody,\r\n .dlx-table tr,\r\n .dlx-table td{\r\n display:block;\r\n width:100%;\r\n }\r\n\r\n .dlx-table tr{\r\n border-bottom:1px solid var(--dlx-border);\r\n padding:10px 12px;\r\n }\r\n\r\n .dlx-table tbody td{\r\n border:0;\r\n padding:8px 0;\r\n background:transparent !important;\r\n }\r\n\r\n .dlx-table tbody td::before{\r\n content: attr(data-label);\r\n display:block;\r\n font-weight:800;\r\n font-size:12px;\r\n color: rgba(15,23,42,.70);\r\n margin-bottom:4px;\r\n }\r\n }\r\n\r\n \/* =======================\r\n Mermaid blocks\r\n ======================= *\/\r\n .dlx-mermaid{\r\n margin: 0;\r\n padding: 16px;\r\n border: 1px solid var(--dlx-border);\r\n border-radius: var(--dlx-radius-sm);\r\n background: #ffffff;\r\n box-shadow: var(--dlx-shadow-sm);\r\n overflow-x: auto;\r\n }\r\n .dlx-mermaid svg{ max-width:100%; height:auto; }\r\n\r\n \/* =======================\r\n Infographic\r\n ======================= *\/\r\n .dlx-infographic{\r\n margin: 12px 0 0;\r\n border:1px solid var(--dlx-border);\r\n border-radius: var(--dlx-radius);\r\n background:#fff;\r\n box-shadow: var(--dlx-shadow);\r\n }\r\n .dlx-infographic__frame{\r\n overflow:auto;\r\n -webkit-overflow-scrolling:touch;\r\n padding: 10px;\r\n }\r\n\r\n \/* =======================\r\n Reveal animation\r\n ======================= *\/\r\n .dlx-reveal{\r\n opacity:0;\r\n transform: translateY(8px);\r\n transition: opacity .5s ease, transform .5s ease;\r\n }\r\n .dlx-reveal.dlx-is-visible{\r\n opacity:1;\r\n transform: translateY(0);\r\n }\r\n\r\n \/* Footer *\/\r\n footer.dlx-section{\r\n padding-top: 10px;\r\n border-top: 1px dashed rgba(15,23,42,.18);\r\n }\r\n <\/style>\r\n<\/head>\r\n\r\n<body>\r\n\r\n<article\r\n class=\"dlx-article dlx-article--colorful\"\r\n itemscope\r\n itemtype=\"https:\/\/schema.org\/Article\"\r\n style=\"\r\n --dlx-bg:#ffffff;\r\n --dlx-surface:#f8fafc;\r\n --dlx-surface-2:#eef2ff;\r\n --dlx-accent:#0f766e;\r\n --dlx-accent-2:#8b5cf6;\r\n --dlx-accent-soft:rgba(139,92,246,.12);\r\n \"\r\n>\r\n <header class=\"dlx-article__hero\">\r\n <p class=\"dlx-article__eyebrow\">Government AI \u2022 Contracts \u2022 Governance<\/p>\r\n\r\n <h1 itemprop=\"headline\">\r\n OpenAI Pentagon Deal: Contracts, Surveillance Fears, and the \u201cCancel ChatGPT\u201d Backlash\r\n <\/h1>\r\n\r\n <p class=\"dlx-article__lead\" itemprop=\"description\">\r\n A structured, source-driven timeline of defense\/government engagements around OpenAI and Anthropic,\r\n the dispute over \u201clawful use\u201d terms, and what the short-horizon backlash metrics actually do (and don\u2019t) prove.\r\n <\/p>\r\n\r\n <div class=\"dlx-meta\" aria-label=\"Informations sur l\u2019article\">\r\n <span itemprop=\"author\" itemscope itemtype=\"https:\/\/schema.org\/Person\">\r\n <strong>Auteur :<\/strong>\r\n <span itemprop=\"name\">DAILLAC<\/span>\r\n <\/span>\r\n <span>\r\n <strong>Temps de lecture :<\/strong> ~12 min\r\n <\/span>\r\n <\/div>\r\n <\/header>\r\n\r\n <nav class=\"dlx-toc\" aria-label=\"Table des mati\u00e8res\">\r\n <div class=\"dlx-toc__title\">Dans cet article<\/div>\r\n <ul>\r\n <li><a href=\"#executive-summary\">Executive Summary<\/a><\/li>\r\n <li><a href=\"#timeline-of-events\">Timeline of Events<\/a><\/li>\r\n <li><a href=\"#timeline-mermaid\">Mermaid Timeline<\/a><\/li>\r\n <li><a href=\"#key-actors\">Key Actors and Statements<\/a><\/li>\r\n <li><a href=\"#entity-mermaid\">Mermaid Entity Diagram<\/a><\/li>\r\n <li><a href=\"#technical-ethical-issues\">Technical and Ethical Issues<\/a><\/li>\r\n <li><a href=\"#backlash-evidence\">Evidence of Backlash & Migration Signals<\/a><\/li>\r\n <li><a href=\"#business-implications\">Business Implications<\/a><\/li>\r\n <li><a href=\"#sources\">Prioritized Sources and Links<\/a><\/li>\r\n <li><a href=\"#conclusion\">Conclusion<\/a><\/li>\r\n <\/ul>\r\n <\/nav>\r\n\r\n <section id=\"executive-summary\" class=\"dlx-section dlx-reveal\" data-dlx=\"reveal\" itemprop=\"articleBody\">\r\n <h2>Executive Summary<\/h2>\r\n\r\n <div class=\"dlx-callout\">\r\n <div class=\"dlx-callout__title\">What the \u201cOpenAI Pentagon deal\u201d actually is<\/div>\r\n <p>\r\n The \u201cOpenAI Pentagon deal\u201d is best understood as a <strong>stack of related U.S. defense and government engagements<\/strong> rather than a single event:\r\n a <strong>$200M ceiling<\/strong> prototype agreement (June 2025) with the U.S. defense establishment; a deployment of a\r\n <strong>customized ChatGPT instance<\/strong> to the DoD\u2019s <strong>GenAI.mil<\/strong> for <strong>unclassified work<\/strong> (Feb 2026); and a separate agreement for deploying OpenAI systems into\r\n <strong>classified environments<\/strong> (late Feb 2026)\u2014followed by a <strong>contractual amendment (Mar 2, 2026)<\/strong> clarifying a domestic-surveillance prohibition.\r\n <\/p>\r\n <\/div>\r\n\r\n <div class=\"dlx-grid dlx-grid--2\">\r\n <div class=\"dlx-note\">\r\n <div class=\"dlx-note__title\">Why it blew up fast<\/div>\r\n <p>\r\n The controversy escalated rapidly because OpenAI and Anthropic were negotiating with the U.S. defense establishment under contested \u201cany lawful use\u201d language.\r\n Anthropic\u2019s CEO <strong>Dario Amodei<\/strong> publicly refused expanded terms that, in his view, could enable <strong>mass domestic surveillance<\/strong> and\r\n <strong>fully autonomous weapons<\/strong>\u2014and described threats of a <strong>\u201csupply chain risk\u201d<\/strong> label and potential use of the\r\n <strong>Defense Production Act<\/strong>.\r\n <\/p>\r\n <\/div>\r\n\r\n <div class=\"dlx-note\">\r\n <div class=\"dlx-note__title\">OpenAI\u2019s stated \u201cred lines\u201d<\/div>\r\n <ul>\r\n <li><strong>No mass domestic surveillance<\/strong><\/li>\r\n <li><strong>No directing autonomous weapons systems<\/strong><\/li>\r\n <li><strong>No high-stakes automated decisions<\/strong> (e.g., \u201csocial credit\u201d)<\/li>\r\n <\/ul>\r\n <\/div>\r\n <\/div>\r\n\r\n <div class=\"dlx-callout\">\r\n <div class=\"dlx-callout__title\">Buyer takeaway (enterprise)<\/div>\r\n <p class=\"dlx-mb-0\">\r\n For enterprise buyers such as <strong>Daillac<\/strong>, this is less about partisan politics than about\r\n <strong>vendor governance<\/strong>, <strong>contract language<\/strong>, <strong>deployment architecture<\/strong>, <strong>data controls<\/strong>, and <strong>resilience<\/strong>.\r\n Your risk is not only \u201cmodel quality,\u201d but also <strong>policy shocks<\/strong>, <strong>reputational spillover<\/strong>, and <strong>supply chain concentration<\/strong>.\r\n <\/p>\r\n <\/div>\r\n <\/section>\r\n\r\n <section id=\"timeline-of-events\" class=\"dlx-section dlx-reveal\" data-dlx=\"reveal\">\r\n <h2>Timeline of Events<\/h2>\r\n <p class=\"dlx-muted\">\r\n The table below separates <strong>event dates<\/strong> (what happened) from <strong>publication dates<\/strong> (when the information surfaced),\r\n because much of the controversy unfolded within ~72 hours.\r\n <\/p>\r\n\r\n \r\n <div class=\"dlx-table-wrap dlx-table-wrap--timeline\" role=\"region\" aria-label=\"Timeline table\" tabindex=\"0\">\r\n <table class=\"dlx-table\" aria-describedby=\"timeline-table-caption\">\r\n <caption id=\"timeline-table-caption\" class=\"dlx-sr-only\">\r\n Timeline of events with date, what happened, why it mattered, and sources reference.\r\n <\/caption>\r\n\r\n <colgroup>\r\n <col style=\"width:14%\">\r\n <col style=\"width:36%\">\r\n <col style=\"width:28%\">\r\n <col style=\"width:22%\">\r\n <\/colgroup>\r\n\r\n <thead>\r\n <tr>\r\n <th scope=\"col\">Date (event)<\/th>\r\n <th scope=\"col\">What happened (high-level)<\/th>\r\n <th scope=\"col\">Why it mattered<\/th>\r\n <th scope=\"col\">Key primary\/credible sources<\/th>\r\n <\/tr>\r\n <\/thead>\r\n\r\n <tbody>\r\n <tr>\r\n <td data-label=\"Date (event)\">Feb 24, 2020<\/td>\r\n <td data-label=\"What happened (high-level)\">DoD adopts AI ethical principles (Responsible, Equitable, Traceable, Reliable, Governable).<\/td>\r\n <td data-label=\"Why it mattered\">Establishes official framing for \u201cresponsible AI\u201d in defense contexts.<\/td>\r\n <td data-label=\"Key sources\">See prioritized sources section.<\/td>\r\n <\/tr>\r\n\r\n <tr>\r\n <td data-label=\"Date (event)\">Jan 25, 2023<\/td>\r\n <td data-label=\"What happened (high-level)\">DoD updates Directive 3000.09 on autonomy in weapon systems (testing discipline and \u201cappropriate levels of human judgment over the use of force\u201d).<\/td>\r\n <td data-label=\"Why it mattered\">Becomes a policy backbone repeatedly referenced in later AI-in-defense debates.<\/td>\r\n <td data-label=\"Key sources\">See prioritized sources section.<\/td>\r\n <\/tr>\r\n\r\n <tr>\r\n <td data-label=\"Date (event)\">Jan 17, 2024<\/td>\r\n <td data-label=\"What happened (high-level)\">OpenAI publicly discusses Pentagon collaboration areas (e.g., cybersecurity) while emphasizing policies still prohibit developing\/using weapons.<\/td>\r\n <td data-label=\"Why it mattered\">Sets the \u201cpre-deal\u201d baseline: bounded, non-weaponized framing.<\/td>\r\n <td data-label=\"Key sources\">See prioritized sources section.<\/td>\r\n <\/tr>\r\n\r\n <tr>\r\n <td data-label=\"Date (event)\">Dec 4, 2024<\/td>\r\n <td data-label=\"What happened (high-level)\">OpenAI partners with Anduril on counter-drone defense use cases; reporting notes policy shifts around \u201cmilitary\/welfare\u201d language.<\/td>\r\n <td data-label=\"Why it mattered\">Shows \u201cdefense\u201d work existed before the 2025\u20132026 contracts.<\/td>\r\n <td data-label=\"Key sources\">See prioritized sources section.<\/td>\r\n <\/tr>\r\n\r\n <tr>\r\n <td data-label=\"Date (event)\">Jun 16, 2025<\/td>\r\n <td data-label=\"What happened (high-level)\">OpenAI launches \u201cOpenAI for Government\u201d and describes a DoD CDAO pilot with a $200M ceiling focused on prototyping.<\/td>\r\n <td data-label=\"Why it mattered\">Formalizes public-sector GTM; introduces \u201ccustom models for national security\u201d on a limited basis.<\/td>\r\n <td data-label=\"Key sources\">See prioritized sources section.<\/td>\r\n <\/tr>\r\n\r\n <tr>\r\n <td data-label=\"Date (event)\">Jun 16, 2025<\/td>\r\n <td data-label=\"What happened (high-level)\">DoD contract award: \u201cOpenAI Public Sector LLC\u201d receives a $200,000,000 fixed-amount prototype OTA; completion estimated July 2026.<\/td>\r\n <td data-label=\"Why it mattered\">The most procurement-literal meaning of \u201cOpenAI Pentagon deal.\u201d<\/td>\r\n <td data-label=\"Key sources\">See prioritized sources section.<\/td>\r\n <\/tr>\r\n\r\n <tr>\r\n <td data-label=\"Date (event)\">Jul 14, 2025<\/td>\r\n <td data-label=\"What happened (high-level)\">Anthropic receives a $200M ceiling DoD prototype OTA; mentions prototypes and fine-tuning on DoD data.<\/td>\r\n <td data-label=\"Why it mattered\">Confirms multi-vendor frontier-lab involvement (not OpenAI-only).<\/td>\r\n <td data-label=\"Key sources\">See prioritized sources section.<\/td>\r\n <\/tr>\r\n\r\n <tr>\r\n <td data-label=\"Date (event)\">Dec 9, 2025<\/td>\r\n <td data-label=\"What happened (high-level)\">DoD launches GenAI.mil; reporting says it integrated Google Gemini and served DoD personnel via the platform.<\/td>\r\n <td data-label=\"Why it mattered\">Substrate for later custom ChatGPT deployment.<\/td>\r\n <td data-label=\"Key sources\">See prioritized sources section.<\/td>\r\n <\/tr>\r\n\r\n <tr>\r\n <td data-label=\"Date (event)\">Feb 9, 2026<\/td>\r\n <td data-label=\"What happened (high-level)\">OpenAI announces custom ChatGPT deployment on GenAI.mil for unclassified work; says data stays isolated and not used for public model training.<\/td>\r\n <td data-label=\"Why it mattered\">Clarifies the unclassified enterprise track.<\/td>\r\n <td data-label=\"Key sources\">See prioritized sources section.<\/td>\r\n <\/tr>\r\n\r\n <tr>\r\n <td data-label=\"Date (event)\">Feb 26\u201327, 2026<\/td>\r\n <td data-label=\"What happened (high-level)\">Anthropic\u2019s CEO publishes refusal: no mass domestic surveillance and no fully autonomous weapons; reporting describes pressure and activism.<\/td>\r\n <td data-label=\"Why it mattered\">Establishes \u201cred lines\u201d that power boycott narratives.<\/td>\r\n <td data-label=\"Key sources\">See prioritized sources section.<\/td>\r\n <\/tr>\r\n\r\n <tr>\r\n <td data-label=\"Date (event)\">Feb 28, 2026<\/td>\r\n <td data-label=\"What happened (high-level)\">OpenAI announces classified-environment agreement (cloud-only, safety stack, cleared personnel); critics focus on \u201cany lawful use.\u201d<\/td>\r\n <td data-label=\"Why it mattered\">Triggers \u201cCancel ChatGPT \/ boycott\u201d wave.<\/td>\r\n <td data-label=\"Key sources\">See prioritized sources section.<\/td>\r\n <\/tr>\r\n\r\n <tr>\r\n <td data-label=\"Date (event)\">Mar 2, 2026<\/td>\r\n <td data-label=\"What happened (high-level)\">OpenAI publishes update adding explicit anti-domestic-surveillance language + clarity around commercially acquired PII; says NSA use would require a new agreement.<\/td>\r\n <td data-label=\"Why it mattered\">Post-backlash tightening; sufficiency still debated.<\/td>\r\n <td data-label=\"Key sources\">See prioritized sources section.<\/td>\r\n <\/tr>\r\n\r\n <tr>\r\n <td data-label=\"Date (event)\">Mar 2\u20134, 2026<\/td>\r\n <td data-label=\"What happened (high-level)\">Analytics show sharp short-term consumer response (uninstalls, reviews) and competitor gains; Reuters reports NATO exploration.<\/td>\r\n <td data-label=\"Why it mattered\">Suggests reputational + commercial feedback loops plus continued expansion.<\/td>\r\n <td data-label=\"Key sources\">See prioritized sources section.<\/td>\r\n <\/tr>\r\n <\/tbody>\r\n <\/table>\r\n <\/div>\r\n <\/section>\r\n\r\n <section id=\"timeline-mermaid\" class=\"dlx-section dlx-reveal\" data-dlx=\"reveal\">\r\n \r\n <\/section>\r\n\r\n <section id=\"key-actors\" class=\"dlx-section dlx-reveal\" data-dlx=\"reveal\">\r\n <h2>Key Actors and Statements<\/h2>\r\n\r\n <div class=\"dlx-card\">\r\n <p>\r\n OpenAI\u2019s position is that its classified-environment agreement has <strong>more guardrails<\/strong> than prior agreements\u2014and that it can enforce red lines through\r\n <strong>deployment architecture (cloud-only)<\/strong>, retained control over its <strong>safety stack<\/strong>, <strong>cleared personnel<\/strong> in the loop, and <strong>contractual remedies<\/strong>\r\n (including termination if the counterparty violates terms).\r\n <\/p>\r\n <p>\r\n The U.S. defense side is visible primarily through procurement records and official policy materials, with many day-to-day negotiation details coming from reporting.\r\n Procurement postings state that \u201cOpenAI Public Sector LLC\u201d was awarded a <strong>$200M OTA<\/strong> to prototype frontier-AI capabilities for \u201cwarfighting and enterprise domains,\u201d\r\n with an estimated completion date of <strong>July 2026<\/strong>.\r\n <\/p>\r\n <p class=\"dlx-mb-0\">\r\n Anthropic\u2019s CEO statement draws two explicit boundaries: <strong>(1) no mass domestic surveillance<\/strong>, and <strong>(2) no fully autonomous weapons<\/strong>.\r\n It also claims the government insisted on \u201cany lawful use\u201d terms and threatened punitive procurement actions.\r\n <\/p>\r\n <\/div>\r\n <\/section>\r\n\r\n <section id=\"entity-mermaid\" class=\"dlx-section dlx-reveal\" data-dlx=\"reveal\">\r\n \r\n <\/section>\r\n\r\n <section id=\"technical-ethical-issues\" class=\"dlx-section dlx-reveal\" data-dlx=\"reveal\">\r\n <h2>Technical and Ethical Issues<\/h2>\r\n\r\n <h3>1) Constraining capability through architecture<\/h3>\r\n <p>\r\n OpenAI\u2019s published terms put heavy weight on <strong>constraining capability through architecture<\/strong>\u2014not just policy. The agreement describes <strong>cloud-only deployment<\/strong>,\r\n explicitly rejecting \u201cguardrails off\u201d models and declining edge deployments.\r\n <\/p>\r\n\r\n <h3>2) Safety stack + classifiers + human access<\/h3>\r\n <p>\r\n OpenAI says the architecture enables independent verification that red lines are not crossed, including running and updating classifiers,\r\n and that cleared OpenAI engineers (and safety\/alignment researchers) will be \u201cin the loop\u201d supporting deployments.\r\n <\/p>\r\n\r\n <div class=\"dlx-note\">\r\n <div class=\"dlx-note__title\">Procurement questions (practical)<\/div>\r\n <ul>\r\n <li>What signals do classifiers monitor, and what are FP\/FN rates?<\/li>\r\n <li>What audit logs exist, who can access them, and what\u2019s retained?<\/li>\r\n <li>Who can override blocks, under what process, and how is it reviewed?<\/li>\r\n <li>How does incident response work in classified environments?<\/li>\r\n <\/ul>\r\n <\/div>\r\n\r\n <h3>3) Surveillance (most contested)<\/h3>\r\n <p>\r\n Anthropic argues \u201cmass domestic surveillance\u201d risks are amplified by AI\u2019s ability to assemble\/interpret fragmented datasets, including data acquired from brokers.\r\n OpenAI\u2019s update (Mar 2, 2026) attempts to close part of the gap by prohibiting intentional domestic surveillance of U.S. persons and clarifying it includes\r\n use of commercially acquired personal\/identifiable information.\r\n <\/p>\r\n\r\n <h3>4) Autonomous weapons<\/h3>\r\n <p>\r\n DoD policy (Directive 3000.09) emphasizes rigorous verification\/validation\/testing and \u201cappropriate levels of human judgment\u201d in the use of force.\r\n OpenAI\u2019s published excerpt says its AI system \u201cwill not be used to independently direct autonomous weapons\u201d where policy requires human control.\r\n Anthropic\u2019s CEO goes further, arguing fully autonomous weapons are not safely supportable today with current frontier systems.\r\n <\/p>\r\n\r\n <h3>5) Data handling and fine-tuning<\/h3>\r\n <p>\r\n OpenAI says GenAI.mil data stays isolated and is not used to train\/improve public or commercial models.\r\n In enterprise contexts, OpenAI positions \u201cno training on business data by default,\u201d plus configurable retention (including zero data retention for qualifying orgs).\r\n Anthropic similarly markets enterprise controls and \u201cno training by default,\u201d while its government prototype work references fine-tuning on DoD data.\r\n <\/p>\r\n\r\n <h3>6) Reliability and mission risk<\/h3>\r\n <p>\r\n Reporting indicates officials worry contractual restrictions or policy enforcement could disrupt operations if tools are constrained mid-mission.\r\n For enterprises, treat \u201cusage policies\u201d and \u201ckill switches\u201d as <strong>availability, continuity, and governance<\/strong> requirements\u2014not abstract ethics.\r\n <\/p>\r\n <\/section>\r\n\r\n <section id=\"backlash-evidence\" class=\"dlx-section dlx-reveal\" data-dlx=\"reveal\">\r\n <h2>Evidence of Backlash, Boycott Behavior, and Migration Signals<\/h2>\r\n\r\n <h3>Backlash narrative themes<\/h3>\r\n <p>\r\n The backlash narrative clustered around: <strong>(1) perceived ethical betrayal<\/strong>, <strong>(2) skepticism about \u201call lawful purposes\u201d language<\/strong>,\r\n and <strong>(3) distrust that legal frameworks sufficiently prevent AI-enabled surveillance<\/strong>.\r\n <\/p>\r\n\r\n <h3>Measured short-horizon indicators (proxies, not churn)<\/h3>\r\n\r\n <div class=\"dlx-grid dlx-grid--4\" aria-label=\"Backlash KPI cards\">\r\n <div class=\"dlx-kpi\">\r\n <span class=\"dlx-kpi__value\" data-dlx-counter=\"295\" data-suffix=\"%\">0%<\/span>\r\n <span class=\"dlx-kpi__label\">ChatGPT uninstalls (US, Feb 28)<\/span>\r\n <\/div>\r\n <div class=\"dlx-kpi\">\r\n <span class=\"dlx-kpi__value\" data-dlx-counter=\"-13\" data-suffix=\"%\">0%<\/span>\r\n <span class=\"dlx-kpi__label\">ChatGPT downloads (US, Feb 28)<\/span>\r\n <\/div>\r\n <div class=\"dlx-kpi\">\r\n <span class=\"dlx-kpi__value\" data-dlx-counter=\"775\" data-suffix=\"%\">0%<\/span>\r\n <span class=\"dlx-kpi__label\">ChatGPT 1-star reviews (Feb 28)<\/span>\r\n <\/div>\r\n <div class=\"dlx-kpi\">\r\n <span class=\"dlx-kpi__value\" data-dlx-counter=\"51\" data-suffix=\"%\">0%<\/span>\r\n <span class=\"dlx-kpi__label\">Claude downloads (US, Feb 28)<\/span>\r\n <\/div>\r\n <\/div>\r\n\r\n <div class=\"dlx-note\">\r\n <div class=\"dlx-note__title\">Interpretation caveat<\/div>\r\n <p class=\"dlx-mb-0\">\r\n These are \u201cshock\u201d metrics (short-horizon proxies), not audited churn counts.\r\n Uninstall \u2260 cancel, and cancel \u2260 stop using web. Treat precise migration numbers without transparent methodology skeptically.\r\n <\/p>\r\n <\/div>\r\n\r\n <h3>Boycott proxy chart and infographic<\/h3>\r\n \r\n\r\n <figure class=\"dlx-infographic\" aria-label=\"Infographic SVG: timeline and proxy metrics\">\r\n <div class=\"dlx-infographic__frame\" role=\"group\" aria-label=\"SVG frame\">\r\n <svg xmlns=\"http:\/\/www.w3.org\/2000\/svg\" width=\"1100\" height=\"650\" viewBox=\"0 0 1100 650\" role=\"img\" aria-label=\"Infographic: OpenAI Pentagon deal timeline and backlash metrics\">\r\n <style>\r\n .title { font: 700 28px Arial, sans-serif; fill: #111; }\r\n .subtitle { font: 600 16px Arial, sans-serif; fill: #111; }\r\n .label { font: 14px Arial, sans-serif; fill: #111; }\r\n .small { font: 12px Arial, sans-serif; fill: #333; }\r\n .box { fill: #f7f7f7; stroke: #222; stroke-width: 1.2; }\r\n .line { stroke: #222; stroke-width: 2; }\r\n .bar { fill: #bdbdbd; stroke: #222; stroke-width: 1; }\r\n .note { font: italic 12px Arial, sans-serif; fill: #333; }\r\n <\/style>\r\n\r\n <text x=\"40\" y=\"55\" class=\"title\">OpenAI \u201cPentagon deal\u201d \u2014 timeline, red lines, and backlash signals<\/text>\r\n <text x=\"40\" y=\"80\" class=\"small\">Data points from public statements and third-party app-market analytics (Feb\u2013Mar 2026). See citations in report text.<\/text>\r\n\r\n <rect x=\"40\" y=\"110\" width=\"500\" height=\"470\" class=\"box\"\/>\r\n <rect x=\"560\" y=\"110\" width=\"500\" height=\"470\" class=\"box\"\/>\r\n\r\n <text x=\"60\" y=\"145\" class=\"subtitle\">Key timeline (contract + deployments)<\/text>\r\n <line x1=\"90\" y1=\"175\" x2=\"90\" y2=\"545\" class=\"line\"\/>\r\n <circle cx=\"90\" cy=\"210\" r=\"8\" fill=\"#fff\" stroke=\"#222\" stroke-width=\"2\"\/>\r\n <circle cx=\"90\" cy=\"305\" r=\"8\" fill=\"#fff\" stroke=\"#222\" stroke-width=\"2\"\/>\r\n <circle cx=\"90\" cy=\"400\" r=\"8\" fill=\"#fff\" stroke=\"#222\" stroke-width=\"2\"\/>\r\n <circle cx=\"90\" cy=\"495\" r=\"8\" fill=\"#fff\" stroke=\"#222\" stroke-width=\"2\"\/>\r\n\r\n <text x=\"115\" y=\"205\" class=\"label\">Jun 16, 2025: $200M ceiling OTA<\/text>\r\n <text x=\"115\" y=\"225\" class=\"small\">Prototype frontier-AI capabilities for warfighting + enterprise domains; completion ~Jul 2026<\/text>\r\n\r\n <text x=\"115\" y=\"300\" class=\"label\">Feb 9, 2026: ChatGPT to GenAI.mil<\/text>\r\n <text x=\"115\" y=\"320\" class=\"small\">Custom ChatGPT for unclassified work; runs in authorized gov cloud; data isolated from public model training<\/text>\r\n\r\n <text x=\"115\" y=\"395\" class=\"label\">Feb 28, 2026: classified deployment agreement<\/text>\r\n <text x=\"115\" y=\"415\" class=\"small\">Cloud-only deployment + OpenAI safety stack; \u201call lawful purposes\u201d language; 3 published \u201cred lines\u201d<\/text>\r\n\r\n <text x=\"115\" y=\"490\" class=\"label\">Mar 2, 2026: amendment on domestic surveillance<\/text>\r\n <text x=\"115\" y=\"510\" class=\"small\">Explicit ban on intentional domestic surveillance of U.S. persons; adds clarity re: commercially acquired PII<\/text>\r\n\r\n <text x=\"580\" y=\"145\" class=\"subtitle\">Backlash proxy metrics (U.S. mobile)<\/text>\r\n <text x=\"580\" y=\"170\" class=\"small\">Day-over-day changes reported around Feb 27\u2013Mar 2, 2026<\/text>\r\n\r\n <text x=\"580\" y=\"210\" class=\"label\">ChatGPT uninstalls (Feb 28)<\/text>\r\n <rect x=\"580\" y=\"220\" width=\"380\" height=\"26\" class=\"bar\"\/>\r\n <text x=\"970\" y=\"239\" class=\"label\">+295%<\/text>\r\n <text x=\"580\" y=\"260\" class=\"small\">Baseline: ~9% typical day-over-day uninstall rate (previous 30 days)<\/text>\r\n\r\n <text x=\"580\" y=\"310\" class=\"label\">ChatGPT downloads (Feb 28)<\/text>\r\n <rect x=\"580\" y=\"320\" width=\"210\" height=\"26\" class=\"bar\"\/>\r\n <text x=\"800\" y=\"339\" class=\"label\">\u221213%<\/text>\r\n <text x=\"580\" y=\"360\" class=\"small\">Prior day (Feb 27) had been +14% day-over-day<\/text>\r\n\r\n <text x=\"580\" y=\"410\" class=\"label\">ChatGPT 1-star reviews (Feb 28)<\/text>\r\n <rect x=\"580\" y=\"420\" width=\"430\" height=\"26\" class=\"bar\"\/>\r\n <text x=\"1020\" y=\"439\" class=\"label\">+775%<\/text>\r\n\r\n <text x=\"580\" y=\"470\" class=\"label\">Claude downloads (Feb 28)<\/text>\r\n <rect x=\"580\" y=\"480\" width=\"250\" height=\"26\" class=\"bar\"\/>\r\n <text x=\"840\" y=\"499\" class=\"label\">+51%<\/text>\r\n\r\n <text x=\"580\" y=\"536\" class=\"note\">Interpretation: short-term sentiment shock; not a full churn estimate.<\/text>\r\n <text x=\"580\" y=\"560\" class=\"small\">Best practice: track retention, cohort conversion, and web traffic over 30\u201390 days.<\/text>\r\n\r\n <text x=\"40\" y=\"620\" class=\"small\">Infographic layout is designed by DAILLAC<\/text>\r\n <\/svg>\r\n <\/div>\r\n <\/figure>\r\n <\/section>\r\n\r\n <section id=\"business-implications\" class=\"dlx-section dlx-reveal\" data-dlx=\"reveal\">\r\n <h2>Business Implications<\/h2>\r\n\r\n <div class=\"dlx-card\">\r\n <p>\r\n The OpenAI Pentagon deal and \u201cChatGPT boycott\u201d wave create a useful stress test for an AI operating model.\r\n Even if you have no defense clients, it shows how quickly a vendor\u2019s posture can become a procurement and reputation issue.\r\n <\/p>\r\n <p class=\"dlx-mb-0\">\r\n Treat LLM providers as <strong>critical third-party dependencies<\/strong> whose constraints can change: by policy update, contract language,\r\n government pressure, or public backlash.\r\n <\/p>\r\n <\/div>\r\n\r\n <h3>Actionable implications (expert buyer)<\/h3>\r\n\r\n <div class=\"dlx-grid dlx-grid--2\">\r\n <div class=\"dlx-callout\">\r\n <div class=\"dlx-callout__title\">Adopt a multi-model architecture (warm standby)<\/div>\r\n <p class=\"dlx-mb-0\">\r\n Abstract model calls, keep prompts portable, and test fallbacks. Vendor access can be politicized or constrained quickly.\r\n <\/p>\r\n <\/div>\r\n\r\n <div class=\"dlx-callout\">\r\n <div class=\"dlx-callout__title\">Review contract + deployment architecture (not just model quality)<\/div>\r\n <p class=\"dlx-mb-0\">\r\n \u201cWhere the model runs\u201d and \u201cwho controls enforcement\u201d are first-class risk drivers (availability, auditability, termination clauses).\r\n <\/p>\r\n <\/div>\r\n\r\n <div class=\"dlx-callout\">\r\n <div class=\"dlx-callout__title\">Codify internal red lines (surveillance & autonomy)<\/div>\r\n <p class=\"dlx-mb-0\">\r\n Don\u2019t rely on vendor policies alone. Define what you won\u2019t build regardless of what might be \u201clawful\u201d or contractually allowed.\r\n <\/p>\r\n <\/div>\r\n\r\n <div class=\"dlx-callout\">\r\n <div class=\"dlx-callout__title\">Strengthen data classification & retention controls<\/div>\r\n <p class=\"dlx-mb-0\">\r\n Map provider retention\/training defaults to your SOPs, especially for sensitive workflows and regulated datasets.\r\n <\/p>\r\n <\/div>\r\n <\/div>\r\n <\/section>\r\n\r\n <section id=\"sources\" class=\"dlx-section dlx-reveal\" data-dlx=\"reveal\">\r\n <h2>Prioritized Sources and Links<\/h2>\r\n\r\n <h3>Primary sources (contracts \/ official statements)<\/h3>\r\n <ul>\r\n <li><a href=\"https:\/\/openai.com\/index\/our-agreement-with-the-department-of-war\/\" target=\"_blank\" rel=\"noopener\">OpenAI \u2014 \u201cOur agreement with the Department of War\u201d (includes Mar 2, 2026 update + red lines + contract excerpts)<\/a><\/li>\r\n <li><a href=\"https:\/\/openai.com\/index\/bringing-chatgpt-to-genaimil\/\" target=\"_blank\" rel=\"noopener\">OpenAI \u2014 \u201cBringing ChatGPT to GenAI.mil\u201d (unclassified deployment; data isolation claim)<\/a><\/li>\r\n <li><a href=\"https:\/\/openai.com\/global-affairs\/introducing-openai-for-government\/\" target=\"_blank\" rel=\"noopener\">OpenAI \u2014 \u201cIntroducing OpenAI for Government\u201d (program framing; pilot; $200M ceiling)<\/a><\/li>\r\n <li><a href=\"https:\/\/www.anthropic.com\/news\/statement-department-of-war\" target=\"_blank\" rel=\"noopener\">Anthropic CEO statement on DoW negotiations (mass domestic surveillance + fully autonomous weapons red lines)<\/a><\/li>\r\n <li><a href=\"https:\/\/www.anthropic.com\/news\/anthropic-and-the-department-of-defense-to-advance-responsible-ai-in-defense-operations\" target=\"_blank\" rel=\"noopener\">Anthropic DoD agreement announcement ($200M; fine-tuning on DoD data)<\/a><\/li>\r\n <li><a href=\"https:\/\/www.esd.whs.mil\/portals\/54\/documents\/dd\/issuances\/dodd\/300009p.pdf\" target=\"_blank\" rel=\"noopener\">DoD Directive 3000.09 (PDF)<\/a><\/li>\r\n <li><a href=\"https:\/\/www.war.gov\/News\/Releases\/Release\/Article\/2091996\/dod-adopts-ethical-principles-for-artificial-intelligence\/\" target=\"_blank\" rel=\"noopener\">DoD press release adopting AI Ethical Principles (2020)<\/a><\/li>\r\n <\/ul>\r\n\r\n <h3>High-quality independent reporting \/ analysis (triangulation)<\/h3>\r\n <ul>\r\n <li><a href=\"https:\/\/www.reuters.com\/business\/media-telecom\/openai-details-layered-protections-us-defense-department-pact-2026-02-28\/\" target=\"_blank\" rel=\"noopener\">Reuters \u2014 deal safeguards<\/a><\/li>\r\n <li><a href=\"https:\/\/www.reuters.com\/business\/ai-contract-restrictions-could-threaten-military-missions-us-official-says-2026-03-03\/\" target=\"_blank\" rel=\"noopener\">Reuters \u2014 contract restrictions vs. mission risk; NATO exploration reported<\/a><\/li>\r\n <li><a href=\"https:\/\/www.theverge.com\/ai-artificial-intelligence\/887309\/openai-anthropic-dod-military-pentagon-contract-sam-altman-hegseth\" target=\"_blank\" rel=\"noopener\">The Verge \u2014 surveillance implications and contract framing<\/a><\/li>\r\n <li><a href=\"https:\/\/www.techradar.com\/ai-platforms-assistants\/chatgpt\/no-ethics-at-all-the-cancel-chatgpt-trend-is-growing-after-openai-signs-a-deal-with-the-us-military\" target=\"_blank\" rel=\"noopener\">TechRadar \u2014 \u201cCancel ChatGPT\u201d trend narrative and social proof<\/a><\/li>\r\n <li><a href=\"https:\/\/www.axios.com\/2026\/02\/27\/pentagon-openai-safety-red-lines-anthropic\" target=\"_blank\" rel=\"noopener\">Axios \u2014 safety concerns in OpenAI-Pentagon terms<\/a><\/li>\r\n <li><a href=\"https:\/\/www.theguardian.com\/us-news\/2026\/feb\/27\/trump-anthropic-ai-federal-agencies\" target=\"_blank\" rel=\"noopener\">The Guardian \u2014 policy conflict and contract amendment coverage<\/a><\/li>\r\n <li><a href=\"https:\/\/techcrunch.com\/2026\/02\/27\/employees-at-google-and-openai-support-anthropics-pentagon-stand-in-open-letter\/\" target=\"_blank\" rel=\"noopener\">TechCrunch \u2014 uninstall\/download\/review metrics and employee open letter<\/a><\/li>\r\n <\/ul>\r\n\r\n <h3>Civil society + research \/ watchdog<\/h3>\r\n <ul>\r\n <li><a href=\"https:\/\/www.eff.org\/deeplinks\/2026\/03\/anthropic-dod-conflict-privacy-protections-shouldnt-depend-decisions-few-powerful\" target=\"_blank\" rel=\"noopener\">Electronic Frontier Foundation \u2014 analysis<\/a><\/li>\r\n <li><a href=\"https:\/\/ainowinstitute.org\/wp-content\/uploads\/2024\/10\/Mind_the_Gap__Foundation_Models_and_the_Covert_Proliferation_of_Military_Intelligence__Surveillance__and_Targeting.pdf\" target=\"_blank\" rel=\"noopener\">AI Now Institute \u2014 report (PDF)<\/a><\/li>\r\n <\/ul>\r\n\r\n <h3>Consulting\/governance frameworks (enterprise operating model)<\/h3>\r\n <ul>\r\n <li><a href=\"https:\/\/www.mckinsey.com\/capabilities\/risk-and-resilience\/our-insights\/implementing-generative-ai-with-speed-and-safety\" target=\"_blank\" rel=\"noopener\">McKinsey \u2014 implementing gen-AI with speed and safety<\/a><\/li>\r\n <li><a href=\"https:\/\/www.bcg.com\/publications\/2023\/a-guide-to-mitigating-ai-risks?utm_source=chatgpt.com\" target=\"_blank\" rel=\"noopener\">Boston Consulting Group \u2014 mitigating AI risks<\/a><\/li>\r\n <li><a href=\"https:\/\/www.bain.com\/insights\/how-control-functions-can-enable-ai-ambition-at-scale\/?utm_source=chatgpt.com\" target=\"_blank\" rel=\"noopener\">Bain \u2014 control functions & vendor risks<\/a><\/li>\r\n <\/ul>\r\n\r\n <h3>Product\/privacy and pricing references<\/h3>\r\n <ul>\r\n <li><a href=\"https:\/\/openai.com\/enterprise-privacy\/\" target=\"_blank\" rel=\"noopener\">OpenAI enterprise privacy<\/a><\/li>\r\n <li><a href=\"https:\/\/openai.com\/business-data\/\" target=\"_blank\" rel=\"noopener\">OpenAI business data (\u201cYour Data\u201d)<\/a><\/li>\r\n <li><a href=\"https:\/\/openai.com\/api\/pricing\/?utm_source=chatgpt.com\" target=\"_blank\" rel=\"noopener\">OpenAI API pricing<\/a><\/li>\r\n <li><a href=\"https:\/\/www.anthropic.com\/pricing\" target=\"_blank\" rel=\"noopener\">Anthropic pricing<\/a><\/li>\r\n <li><a href=\"https:\/\/docs.anthropic.com\/en\/docs\/about-claude\/pricing\" target=\"_blank\" rel=\"noopener\">Anthropic API pricing<\/a><\/li>\r\n <li><a href=\"https:\/\/ai.google.dev\/gemini-api\/docs\/pricing?utm_source=chatgpt.com\" target=\"_blank\" rel=\"noopener\">Gemini Developer API pricing<\/a><\/li>\r\n <li><a href=\"https:\/\/mistral.ai\/pricing\" target=\"_blank\" rel=\"noopener\">Mistral pricing<\/a><\/li>\r\n <\/ul>\r\n\r\n <h3>Additional (French) coverage consulted<\/h3>\r\n <ul>\r\n <li><a href=\"https:\/\/www.usine-digitale.fr\/article\/openai-decroche-un-contrat-a-200-millions-de-dollars-avec-le-pentagone.N2233872\" target=\"_blank\" rel=\"noopener\">L\u2019Usine Digitale \u2014 2025 contract framing (French)<\/a><\/li>\r\n <li><a href=\"https:\/\/www.frenchweb.fr\/openai-met-un-pied-dans-le-complexe-militaro-tech-americain\/455493\" target=\"_blank\" rel=\"noopener\">FrenchWeb \u2014 commentary (French)<\/a><\/li>\r\n <li><a href=\"https:\/\/www.silicon.fr\/business-1367\/ia-militaire-openai-remplace-anthropic-au-pentagone-225961\" target=\"_blank\" rel=\"noopener\">Silicon.fr \u2014 coverage (French)<\/a><\/li>\r\n <\/ul>\r\n <\/section>\r\n\r\n <footer id=\"conclusion\" class=\"dlx-section dlx-reveal\" data-dlx=\"reveal\">\r\n <h2>Conclusion<\/h2>\r\n <p>\r\n The core story is not a single \u201cdeal,\u201d but a sequence: prototype contracting, unclassified platform deployments,\r\n then a contested shift toward classified environments\u2014followed by rapid contractual clarification under public pressure.\r\n For enterprise governance, the durable lesson is to model <strong>policy volatility<\/strong>, <strong>contract constraints<\/strong>, and <strong>vendor concentration<\/strong>\r\n as first-class operational risks.\r\n <\/p>\r\n <\/footer>\r\n<\/article>\r\n\r\n<script>\r\n\/* ===========================\r\n DAILLAC mini JS (Standalone)\r\n - Reveal on scroll\r\n - Counter animation\r\n - Mermaid render if available\r\n =========================== *\/\r\n(function () {\r\n 'use strict';\r\n\r\n \/\/ ---- Reveal\r\n (function initReveal(){\r\n var nodes = Array.prototype.slice.call(document.querySelectorAll('[data-dlx=\"reveal\"]'));\r\n if (!('IntersectionObserver' in window) || nodes.length === 0) {\r\n nodes.forEach(function(n){ n.classList.add('dlx-is-visible'); });\r\n return;\r\n }\r\n var io = new IntersectionObserver(function(entries){\r\n entries.forEach(function(e){\r\n if (e.isIntersecting) {\r\n e.target.classList.add('dlx-is-visible');\r\n io.unobserve(e.target);\r\n }\r\n });\r\n }, { root: null, threshold: 0.12 });\r\n nodes.forEach(function(n){ io.observe(n); });\r\n })();\r\n\r\n \/\/ ---- Counters\r\n (function initCounters(){\r\n var nodes = Array.prototype.slice.call(document.querySelectorAll('[data-dlx-counter]'));\r\n if (nodes.length === 0) return;\r\n\r\n function animate(el){\r\n if (el.dataset.dlxCounted === 'true') return;\r\n el.dataset.dlxCounted = 'true';\r\n\r\n var target = parseFloat(el.getAttribute('data-dlx-counter') || '0');\r\n var suffix = el.getAttribute('data-suffix') || '';\r\n var duration = 650;\r\n var start = null;\r\n\r\n function frame(ts){\r\n if (!start) start = ts;\r\n var p = Math.min(1, (ts - start) \/ duration);\r\n \/\/ easeOutCubic\r\n var eased = 1 - Math.pow(1 - p, 3);\r\n var value = target * eased;\r\n\r\n \/\/ Format integers cleanly (keep minus sign)\r\n var out = (Math.abs(target) >= 10)\r\n ? Math.round(value).toString()\r\n : value.toFixed(1);\r\n\r\n \/\/ Ensure minus sign stays correct near 0 for negative targets\r\n if (target < 0 && parseFloat(out) > 0) out = '-' + out;\r\n\r\n el.textContent = out + suffix;\r\n if (p < 1) requestAnimationFrame(frame);\r\n else el.textContent = (Math.round(target)).toString() + suffix;\r\n }\r\n requestAnimationFrame(frame);\r\n }\r\n\r\n if (!('IntersectionObserver' in window)) {\r\n nodes.forEach(animate);\r\n return;\r\n }\r\n\r\n var io = new IntersectionObserver(function(entries){\r\n entries.forEach(function(e){\r\n if (e.isIntersecting) {\r\n animate(e.target);\r\n io.unobserve(e.target);\r\n }\r\n });\r\n }, { threshold: 0.35 });\r\n\r\n nodes.forEach(function(n){ io.observe(n); });\r\n })();\r\n\r\n \/\/ ---- Mermaid render (only if Mermaid is already loaded on the site)\r\n (function initMermaid(){\r\n if (!window.mermaid) return;\r\n\r\n function qsa(sel, root){ return Array.prototype.slice.call((root||document).querySelectorAll(sel)); }\r\n\r\n try {\r\n if (typeof window.mermaid.initialize === 'function') {\r\n window.mermaid.initialize({ startOnLoad: false, securityLevel: 'strict', theme: 'neutral' });\r\n }\r\n } catch (e) {}\r\n\r\n qsa('.dlx-article .mermaid').forEach(function (node, idx) {\r\n if (node.dataset.dlxMermaidRendered === 'true') return;\r\n if (node.querySelector('svg')) { node.dataset.dlxMermaidRendered = 'true'; return; }\r\n\r\n var code = (node.textContent || '').trim();\r\n if (!code) return;\r\n\r\n try {\r\n var id = 'dlx-mermaid-' + idx;\r\n window.mermaid.render(id, code, function (svg) {\r\n node.innerHTML = svg;\r\n node.dataset.dlxMermaidRendered = 'true';\r\n });\r\n } catch (e) {\r\n \/\/ Keep code visible if render fails\r\n }\r\n });\r\n })();\r\n\r\n})();\r\n<\/script>\r\n\r\n<\/body>\r\n<\/html>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<\/div>\n\t\t","protected":false},"excerpt":{"rendered":"<p>OpenAI \u201cPentagon deal\u201d \u2014 timeline, red lines, and backlash signals Government AI \u2022 Contracts \u2022 Governance OpenAI Pentagon Deal: Contracts, Surveillance Fears, and the \u201cCancel ChatGPT\u201d Backlash A structured, source-driven timeline of defense\/government engagements around OpenAI and Anthropic, the dispute over \u201clawful use\u201d terms, and what the short-horizon backlash metrics actually do (and don\u2019t) prove. […]<\/p>\n","protected":false},"author":4,"featured_media":12750,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[],"class_list":["post-12749","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-non-classifie"],"_links":{"self":[{"href":"https:\/\/www.daillac.com\/en\/wp-json\/wp\/v2\/posts\/12749","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.daillac.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.daillac.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.daillac.com\/en\/wp-json\/wp\/v2\/users\/4"}],"replies":[{"embeddable":true,"href":"https:\/\/www.daillac.com\/en\/wp-json\/wp\/v2\/comments?post=12749"}],"version-history":[{"count":16,"href":"https:\/\/www.daillac.com\/en\/wp-json\/wp\/v2\/posts\/12749\/revisions"}],"predecessor-version":[{"id":12768,"href":"https:\/\/www.daillac.com\/en\/wp-json\/wp\/v2\/posts\/12749\/revisions\/12768"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.daillac.com\/en\/wp-json\/wp\/v2\/media\/12750"}],"wp:attachment":[{"href":"https:\/\/www.daillac.com\/en\/wp-json\/wp\/v2\/media?parent=12749"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.daillac.com\/en\/wp-json\/wp\/v2\/categories?post=12749"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.daillac.com\/en\/wp-json\/wp\/v2\/tags?post=12749"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}

\n\t\t\t\t\t\t

\n\t\t\t\t\t

\n\t\t\t

\n\t\t\t\t\t\t

\n\t\t\t\t

\n\t\t\t\t\t\r\n\r\n\r\n \r\n \r\n OpenAI \u201cPentagon deal\u201d \u2014 timeline, red lines, and backlash signals<\/title>\r\n\r\n <style>\r\n \/* ===========================\r\n DAILLAC Article (Standalone)\r\n =========================== *\/\r\n\r\n :root{\r\n --dlx-bg:#ffffff;\r\n --dlx-text:#0f172a;\r\n --dlx-muted:rgba(15,23,42,.72);\r\n --dlx-border:rgba(16,24,40,.10);\r\n\r\n --dlx-surface:#f8fafc;\r\n --dlx-surface-2:#eef2ff;\r\n\r\n --dlx-accent:#0f766e;\r\n --dlx-accent-2:#8b5cf6;\r\n --dlx-accent-soft:rgba(139,92,246,.12);\r\n\r\n --dlx-radius:16px;\r\n --dlx-radius-sm:12px;\r\n\r\n --dlx-shadow: 0 10px 24px rgba(16,24,40,.06);\r\n --dlx-shadow-sm: 0 6px 14px rgba(16,24,40,.06);\r\n }\r\n\r\n *{ box-sizing:border-box; }\r\n html,body{ margin:0; padding:0; background:var(--dlx-bg); color:var(--dlx-text); }\r\n body{ font-family:system-ui,-apple-system,Segoe UI,Roboto,Arial,sans-serif; line-height:1.65; }\r\n\r\n a{ color:var(--dlx-accent); text-decoration:none; }\r\n a:hover{ text-decoration:underline; }\r\n\r\n \/* Accessibility helper *\/\r\n .dlx-sr-only{\r\n position:absolute !important;\r\n width:1px;height:1px;\r\n padding:0;margin:-1px;\r\n overflow:hidden;clip:rect(0,0,0,0);\r\n white-space:nowrap;border:0;\r\n }\r\n\r\n \/* Article container *\/\r\n .dlx-article{\r\n max-width: 980px;\r\n margin: 0 auto;\r\n padding: 22px 16px 56px;\r\n background: var(--dlx-bg);\r\n }\r\n\r\n \/* Hero *\/\r\n .dlx-article__hero{\r\n position:relative;\r\n padding: 18px 16px 18px;\r\n border: 1px solid var(--dlx-border);\r\n border-radius: var(--dlx-radius);\r\n background:\r\n radial-gradient(1200px 500px at 0% 0%, rgba(15,118,110,.10), transparent 55%),\r\n radial-gradient(900px 420px at 100% 0%, rgba(139,92,246,.10), transparent 60%),\r\n #fff;\r\n box-shadow: var(--dlx-shadow);\r\n }\r\n\r\n .dlx-article__eyebrow{\r\n display:inline-flex;\r\n gap:8px;\r\n align-items:center;\r\n margin: 0 0 10px;\r\n padding: 6px 10px;\r\n border-radius: 999px;\r\n border: 1px solid rgba(15,118,110,.22);\r\n background: rgba(15,118,110,.06);\r\n color: rgba(15,118,110,.95);\r\n font-weight: 700;\r\n font-size: 12px;\r\n letter-spacing: .02em;\r\n }\r\n\r\n .dlx-article h1{\r\n margin: 0 0 10px;\r\n font-size: 34px;\r\n line-height: 1.15;\r\n letter-spacing: -0.02em;\r\n }\r\n\r\n .dlx-article__lead{\r\n margin: 0 0 14px;\r\n color: var(--dlx-muted);\r\n font-size: 16px;\r\n max-width: 72ch;\r\n }\r\n\r\n .dlx-meta{\r\n display:flex;\r\n flex-wrap:wrap;\r\n gap:10px 14px;\r\n padding-top: 10px;\r\n border-top: 1px dashed rgba(15,23,42,.18);\r\n color: rgba(15,23,42,.72);\r\n font-size: 13px;\r\n }\r\n\r\n \/* TOC *\/\r\n .dlx-toc{\r\n margin: 18px 0 0;\r\n padding: 14px 14px;\r\n border: 1px solid var(--dlx-border);\r\n border-radius: var(--dlx-radius);\r\n background:\r\n linear-gradient(90deg, rgba(139,92,246,.08), rgba(15,118,110,.06)),\r\n #fff;\r\n box-shadow: var(--dlx-shadow-sm);\r\n }\r\n .dlx-toc__title{\r\n font-weight: 800;\r\n margin: 0 0 8px;\r\n color: rgba(15,23,42,.92);\r\n }\r\n .dlx-toc ul{\r\n margin:0;\r\n padding-left: 18px;\r\n }\r\n .dlx-toc li{ margin: 6px 0; }\r\n\r\n \/* Sections *\/\r\n .dlx-section{ margin-top: 22px; }\r\n .dlx-section h2{\r\n margin: 0 0 10px;\r\n font-size: 22px;\r\n letter-spacing: -0.01em;\r\n }\r\n .dlx-section h3{\r\n margin: 16px 0 8px;\r\n font-size: 16px;\r\n }\r\n\r\n .dlx-muted{ color: var(--dlx-muted); }\r\n\r\n \/* Layout helpers *\/\r\n .dlx-grid{ display:grid; gap: 12px; }\r\n .dlx-grid--2{ grid-template-columns: repeat(2, minmax(0,1fr)); }\r\n .dlx-grid--4{ grid-template-columns: repeat(4, minmax(0,1fr)); }\r\n @media (max-width: 860px){ .dlx-grid--4{ grid-template-columns: repeat(2, minmax(0,1fr)); } }\r\n @media (max-width: 640px){ .dlx-grid--2, .dlx-grid--4{ grid-template-columns: 1fr; } }\r\n\r\n \/* Cards \/ callouts *\/\r\n .dlx-card{\r\n border: 1px solid var(--dlx-border);\r\n border-radius: var(--dlx-radius);\r\n background: #fff;\r\n padding: 14px 14px;\r\n box-shadow: var(--dlx-shadow-sm);\r\n }\r\n\r\n .dlx-callout{\r\n border: 1px solid rgba(139,92,246,.22);\r\n border-radius: var(--dlx-radius);\r\n background:\r\n linear-gradient(90deg, rgba(139,92,246,.10), rgba(15,118,110,.06)),\r\n #fff;\r\n padding: 14px 14px;\r\n box-shadow: var(--dlx-shadow-sm);\r\n }\r\n .dlx-callout__title{\r\n font-weight: 900;\r\n margin: 0 0 6px;\r\n color: rgba(15,23,42,.92);\r\n }\r\n\r\n .dlx-note{\r\n border: 1px solid rgba(15,118,110,.20);\r\n border-radius: var(--dlx-radius);\r\n background:\r\n radial-gradient(800px 240px at 0% 0%, rgba(15,118,110,.10), transparent 60%),\r\n #fff;\r\n padding: 14px 14px;\r\n box-shadow: var(--dlx-shadow-sm);\r\n }\r\n .dlx-note__title{\r\n font-weight: 900;\r\n margin: 0 0 6px;\r\n color: rgba(15,23,42,.92);\r\n }\r\n\r\n .dlx-mb-0{ margin-bottom: 0 !important; }\r\n\r\n \/* KPI *\/\r\n .dlx-kpi{\r\n border: 1px solid var(--dlx-border);\r\n border-radius: var(--dlx-radius);\r\n background: #fff;\r\n padding: 12px 12px;\r\n box-shadow: var(--dlx-shadow-sm);\r\n }\r\n .dlx-kpi__value{\r\n display:block;\r\n font-weight: 950;\r\n font-size: 28px;\r\n line-height: 1.05;\r\n letter-spacing: -0.02em;\r\n background: linear-gradient(90deg, var(--dlx-accent), var(--dlx-accent-2));\r\n -webkit-background-clip: text;\r\n background-clip: text;\r\n color: transparent;\r\n margin-bottom: 6px;\r\n }\r\n .dlx-kpi__label{\r\n display:block;\r\n color: var(--dlx-muted);\r\n font-size: 12px;\r\n font-weight: 700;\r\n }\r\n\r\n \/* =======================\r\n TABLE (Timeline Improved)\r\n ======================= *\/\r\n .dlx-table-wrap--timeline{\r\n border:1px solid var(--dlx-border);\r\n border-radius: var(--dlx-radius);\r\n background:#fff;\r\n overflow:auto;\r\n -webkit-overflow-scrolling:touch;\r\n box-shadow: var(--dlx-shadow);\r\n }\r\n\r\n .dlx-table{\r\n width:100%;\r\n border-collapse:separate;\r\n border-spacing:0;\r\n table-layout:fixed;\r\n min-width: 920px;\r\n }\r\n\r\n .dlx-table thead th{\r\n position:sticky;\r\n top:0;\r\n z-index:2;\r\n text-align:left;\r\n padding:12px 12px;\r\n background:\r\n linear-gradient(90deg,\r\n rgba(15,118,110,.10),\r\n rgba(139,92,246,.08)\r\n ),\r\n #ffffff;\r\n border-bottom:1px solid var(--dlx-border);\r\n color:#0f172a;\r\n font-size: 13px;\r\n }\r\n\r\n .dlx-table tbody td{\r\n vertical-align:top;\r\n padding:12px 12px;\r\n border-bottom:1px solid rgba(16,24,40,.08);\r\n color:#0f172a;\r\n font-size: 13px;\r\n overflow-wrap:anywhere;\r\n word-break:break-word;\r\n }\r\n\r\n .dlx-table tbody tr:nth-child(odd) td{\r\n background:rgba(15,118,110,.04);\r\n }\r\n\r\n .dlx-table tbody tr:hover td{\r\n background:rgba(139,92,246,.08);\r\n }\r\n\r\n \/* Mobile: rows become cards *\/\r\n @media (max-width: 720px){\r\n .dlx-table{ min-width: 0; table-layout:auto; }\r\n .dlx-table thead{ display:none; }\r\n\r\n .dlx-table,\r\n .dlx-table tbody,\r\n .dlx-table tr,\r\n .dlx-table td{\r\n display:block;\r\n width:100%;\r\n }\r\n\r\n .dlx-table tr{\r\n border-bottom:1px solid var(--dlx-border);\r\n padding:10px 12px;\r\n }\r\n\r\n .dlx-table tbody td{\r\n border:0;\r\n padding:8px 0;\r\n background:transparent !important;\r\n }\r\n\r\n .dlx-table tbody td::before{\r\n content: attr(data-label);\r\n display:block;\r\n font-weight:800;\r\n font-size:12px;\r\n color: rgba(15,23,42,.70);\r\n margin-bottom:4px;\r\n }\r\n }\r\n\r\n \/* =======================\r\n Mermaid blocks\r\n ======================= *\/\r\n .dlx-mermaid{\r\n margin: 0;\r\n padding: 16px;\r\n border: 1px solid var(--dlx-border);\r\n border-radius: var(--dlx-radius-sm);\r\n background: #ffffff;\r\n box-shadow: var(--dlx-shadow-sm);\r\n overflow-x: auto;\r\n }\r\n .dlx-mermaid svg{ max-width:100%; height:auto; }\r\n\r\n \/* =======================\r\n Infographic\r\n ======================= *\/\r\n .dlx-infographic{\r\n margin: 12px 0 0;\r\n border:1px solid var(--dlx-border);\r\n border-radius: var(--dlx-radius);\r\n background:#fff;\r\n box-shadow: var(--dlx-shadow);\r\n }\r\n .dlx-infographic__frame{\r\n overflow:auto;\r\n -webkit-overflow-scrolling:touch;\r\n padding: 10px;\r\n }\r\n\r\n \/* =======================\r\n Reveal animation\r\n ======================= *\/\r\n .dlx-reveal{\r\n opacity:0;\r\n transform: translateY(8px);\r\n transition: opacity .5s ease, transform .5s ease;\r\n }\r\n .dlx-reveal.dlx-is-visible{\r\n opacity:1;\r\n transform: translateY(0);\r\n }\r\n\r\n \/* Footer *\/\r\n footer.dlx-section{\r\n padding-top: 10px;\r\n border-top: 1px dashed rgba(15,23,42,.18);\r\n }\r\n <\/style>\r\n<\/head>\r\n\r\n<body>\r\n\r\n<article\r\n class=\"dlx-article dlx-article--colorful\"\r\n itemscope\r\n itemtype=\"https:\/\/schema.org\/Article\"\r\n style=\"\r\n --dlx-bg:#ffffff;\r\n --dlx-surface:#f8fafc;\r\n --dlx-surface-2:#eef2ff;\r\n --dlx-accent:#0f766e;\r\n --dlx-accent-2:#8b5cf6;\r\n --dlx-accent-soft:rgba(139,92,246,.12);\r\n \"\r\n>\r\n <header class=\"dlx-article__hero\">\r\n <p class=\"dlx-article__eyebrow\">Government AI \u2022 Contracts \u2022 Governance<\/p>\r\n\r\n <h1 itemprop=\"headline\">\r\n OpenAI Pentagon Deal: Contracts, Surveillance Fears, and the \u201cCancel ChatGPT\u201d Backlash\r\n <\/h1>\r\n\r\n <p class=\"dlx-article__lead\" itemprop=\"description\">\r\n A structured, source-driven timeline of defense\/government engagements around OpenAI and Anthropic,\r\n the dispute over \u201clawful use\u201d terms, and what the short-horizon backlash metrics actually do (and don\u2019t) prove.\r\n <\/p>\r\n\r\n <div class=\"dlx-meta\" aria-label=\"Informations sur l\u2019article\">\r\n <span itemprop=\"author\" itemscope itemtype=\"https:\/\/schema.org\/Person\">\r\n <strong>Auteur :<\/strong>\r\n <span itemprop=\"name\">DAILLAC<\/span>\r\n <\/span>\r\n <span>\r\n <strong>Temps de lecture :<\/strong> ~12 min\r\n <\/span>\r\n <\/div>\r\n <\/header>\r\n\r\n <nav class=\"dlx-toc\" aria-label=\"Table des mati\u00e8res\">\r\n <div class=\"dlx-toc__title\">Dans cet article<\/div>\r\n <ul>\r\n <li><a href=\"#executive-summary\">Executive Summary<\/a><\/li>\r\n <li><a href=\"#timeline-of-events\">Timeline of Events<\/a><\/li>\r\n <li><a href=\"#timeline-mermaid\">Mermaid Timeline<\/a><\/li>\r\n <li><a href=\"#key-actors\">Key Actors and Statements<\/a><\/li>\r\n <li><a href=\"#entity-mermaid\">Mermaid Entity Diagram<\/a><\/li>\r\n <li><a href=\"#technical-ethical-issues\">Technical and Ethical Issues<\/a><\/li>\r\n <li><a href=\"#backlash-evidence\">Evidence of Backlash & Migration Signals<\/a><\/li>\r\n <li><a href=\"#business-implications\">Business Implications<\/a><\/li>\r\n <li><a href=\"#sources\">Prioritized Sources and Links<\/a><\/li>\r\n <li><a href=\"#conclusion\">Conclusion<\/a><\/li>\r\n <\/ul>\r\n <\/nav>\r\n\r\n <section id=\"executive-summary\" class=\"dlx-section dlx-reveal\" data-dlx=\"reveal\" itemprop=\"articleBody\">\r\n <h2>Executive Summary<\/h2>\r\n\r\n <div class=\"dlx-callout\">\r\n <div class=\"dlx-callout__title\">What the \u201cOpenAI Pentagon deal\u201d actually is<\/div>\r\n <p>\r\n The \u201cOpenAI Pentagon deal\u201d is best understood as a <strong>stack of related U.S. defense and government engagements<\/strong> rather than a single event:\r\n a <strong>$200M ceiling<\/strong> prototype agreement (June 2025) with the U.S. defense establishment; a deployment of a\r\n <strong>customized ChatGPT instance<\/strong> to the DoD\u2019s <strong>GenAI.mil<\/strong> for <strong>unclassified work<\/strong> (Feb 2026); and a separate agreement for deploying OpenAI systems into\r\n <strong>classified environments<\/strong> (late Feb 2026)\u2014followed by a <strong>contractual amendment (Mar 2, 2026)<\/strong> clarifying a domestic-surveillance prohibition.\r\n <\/p>\r\n <\/div>\r\n\r\n <div class=\"dlx-grid dlx-grid--2\">\r\n <div class=\"dlx-note\">\r\n <div class=\"dlx-note__title\">Why it blew up fast<\/div>\r\n <p>\r\n The controversy escalated rapidly because OpenAI and Anthropic were negotiating with the U.S. defense establishment under contested \u201cany lawful use\u201d language.\r\n Anthropic\u2019s CEO <strong>Dario Amodei<\/strong> publicly refused expanded terms that, in his view, could enable <strong>mass domestic surveillance<\/strong> and\r\n <strong>fully autonomous weapons<\/strong>\u2014and described threats of a <strong>\u201csupply chain risk\u201d<\/strong> label and potential use of the\r\n <strong>Defense Production Act<\/strong>.\r\n <\/p>\r\n <\/div>\r\n\r\n <div class=\"dlx-note\">\r\n <div class=\"dlx-note__title\">OpenAI\u2019s stated \u201cred lines\u201d<\/div>\r\n <ul>\r\n <li><strong>No mass domestic surveillance<\/strong><\/li>\r\n <li><strong>No directing autonomous weapons systems<\/strong><\/li>\r\n <li><strong>No high-stakes automated decisions<\/strong> (e.g., \u201csocial credit\u201d)<\/li>\r\n <\/ul>\r\n <\/div>\r\n <\/div>\r\n\r\n <div class=\"dlx-callout\">\r\n <div class=\"dlx-callout__title\">Buyer takeaway (enterprise)<\/div>\r\n <p class=\"dlx-mb-0\">\r\n For enterprise buyers such as <strong>Daillac<\/strong>, this is less about partisan politics than about\r\n <strong>vendor governance<\/strong>, <strong>contract language<\/strong>, <strong>deployment architecture<\/strong>, <strong>data controls<\/strong>, and <strong>resilience<\/strong>.\r\n Your risk is not only \u201cmodel quality,\u201d but also <strong>policy shocks<\/strong>, <strong>reputational spillover<\/strong>, and <strong>supply chain concentration<\/strong>.\r\n <\/p>\r\n <\/div>\r\n <\/section>\r\n\r\n <section id=\"timeline-of-events\" class=\"dlx-section dlx-reveal\" data-dlx=\"reveal\">\r\n <h2>Timeline of Events<\/h2>\r\n <p class=\"dlx-muted\">\r\n The table below separates <strong>event dates<\/strong> (what happened) from <strong>publication dates<\/strong> (when the information surfaced),\r\n because much of the controversy unfolded within ~72 hours.\r\n <\/p>\r\n\r\n \r\n <div class=\"dlx-table-wrap dlx-table-wrap--timeline\" role=\"region\" aria-label=\"Timeline table\" tabindex=\"0\">\r\n <table class=\"dlx-table\" aria-describedby=\"timeline-table-caption\">\r\n <caption id=\"timeline-table-caption\" class=\"dlx-sr-only\">\r\n Timeline of events with date, what happened, why it mattered, and sources reference.\r\n <\/caption>\r\n\r\n <colgroup>\r\n <col style=\"width:14%\">\r\n <col style=\"width:36%\">\r\n <col style=\"width:28%\">\r\n <col style=\"width:22%\">\r\n <\/colgroup>\r\n\r\n <thead>\r\n <tr>\r\n <th scope=\"col\">Date (event)<\/th>\r\n <th scope=\"col\">What happened (high-level)<\/th>\r\n <th scope=\"col\">Why it mattered<\/th>\r\n <th scope=\"col\">Key primary\/credible sources<\/th>\r\n <\/tr>\r\n <\/thead>\r\n\r\n <tbody>\r\n <tr>\r\n <td data-label=\"Date (event)\">Feb 24, 2020<\/td>\r\n <td data-label=\"What happened (high-level)\">DoD adopts AI ethical principles (Responsible, Equitable, Traceable, Reliable, Governable).<\/td>\r\n <td data-label=\"Why it mattered\">Establishes official framing for \u201cresponsible AI\u201d in defense contexts.<\/td>\r\n <td data-label=\"Key sources\">See prioritized sources section.<\/td>\r\n <\/tr>\r\n\r\n <tr>\r\n <td data-label=\"Date (event)\">Jan 25, 2023<\/td>\r\n <td data-label=\"What happened (high-level)\">DoD updates Directive 3000.09 on autonomy in weapon systems (testing discipline and \u201cappropriate levels of human judgment over the use of force\u201d).<\/td>\r\n <td data-label=\"Why it mattered\">Becomes a policy backbone repeatedly referenced in later AI-in-defense debates.<\/td>\r\n <td data-label=\"Key sources\">See prioritized sources section.<\/td>\r\n <\/tr>\r\n\r\n <tr>\r\n <td data-label=\"Date (event)\">Jan 17, 2024<\/td>\r\n <td data-label=\"What happened (high-level)\">OpenAI publicly discusses Pentagon collaboration areas (e.g., cybersecurity) while emphasizing policies still prohibit developing\/using weapons.<\/td>\r\n <td data-label=\"Why it mattered\">Sets the \u201cpre-deal\u201d baseline: bounded, non-weaponized framing.<\/td>\r\n <td data-label=\"Key sources\">See prioritized sources section.<\/td>\r\n <\/tr>\r\n\r\n <tr>\r\n <td data-label=\"Date (event)\">Dec 4, 2024<\/td>\r\n <td data-label=\"What happened (high-level)\">OpenAI partners with Anduril on counter-drone defense use cases; reporting notes policy shifts around \u201cmilitary\/welfare\u201d language.<\/td>\r\n <td data-label=\"Why it mattered\">Shows \u201cdefense\u201d work existed before the 2025\u20132026 contracts.<\/td>\r\n <td data-label=\"Key sources\">See prioritized sources section.<\/td>\r\n <\/tr>\r\n\r\n <tr>\r\n <td data-label=\"Date (event)\">Jun 16, 2025<\/td>\r\n <td data-label=\"What happened (high-level)\">OpenAI launches \u201cOpenAI for Government\u201d and describes a DoD CDAO pilot with a $200M ceiling focused on prototyping.<\/td>\r\n <td data-label=\"Why it mattered\">Formalizes public-sector GTM; introduces \u201ccustom models for national security\u201d on a limited basis.<\/td>\r\n <td data-label=\"Key sources\">See prioritized sources section.<\/td>\r\n <\/tr>\r\n\r\n <tr>\r\n <td data-label=\"Date (event)\">Jun 16, 2025<\/td>\r\n <td data-label=\"What happened (high-level)\">DoD contract award: \u201cOpenAI Public Sector LLC\u201d receives a $200,000,000 fixed-amount prototype OTA; completion estimated July 2026.<\/td>\r\n <td data-label=\"Why it mattered\">The most procurement-literal meaning of \u201cOpenAI Pentagon deal.\u201d<\/td>\r\n <td data-label=\"Key sources\">See prioritized sources section.<\/td>\r\n <\/tr>\r\n\r\n <tr>\r\n <td data-label=\"Date (event)\">Jul 14, 2025<\/td>\r\n <td data-label=\"What happened (high-level)\">Anthropic receives a $200M ceiling DoD prototype OTA; mentions prototypes and fine-tuning on DoD data.<\/td>\r\n <td data-label=\"Why it mattered\">Confirms multi-vendor frontier-lab involvement (not OpenAI-only).<\/td>\r\n <td data-label=\"Key sources\">See prioritized sources section.<\/td>\r\n <\/tr>\r\n\r\n <tr>\r\n <td data-label=\"Date (event)\">Dec 9, 2025<\/td>\r\n <td data-label=\"What happened (high-level)\">DoD launches GenAI.mil; reporting says it integrated Google Gemini and served DoD personnel via the platform.<\/td>\r\n <td data-label=\"Why it mattered\">Substrate for later custom ChatGPT deployment.<\/td>\r\n <td data-label=\"Key sources\">See prioritized sources section.<\/td>\r\n <\/tr>\r\n\r\n <tr>\r\n <td data-label=\"Date (event)\">Feb 9, 2026<\/td>\r\n <td data-label=\"What happened (high-level)\">OpenAI announces custom ChatGPT deployment on GenAI.mil for unclassified work; says data stays isolated and not used for public model training.<\/td>\r\n <td data-label=\"Why it mattered\">Clarifies the unclassified enterprise track.<\/td>\r\n <td data-label=\"Key sources\">See prioritized sources section.<\/td>\r\n <\/tr>\r\n\r\n <tr>\r\n <td data-label=\"Date (event)\">Feb 26\u201327, 2026<\/td>\r\n <td data-label=\"What happened (high-level)\">Anthropic\u2019s CEO publishes refusal: no mass domestic surveillance and no fully autonomous weapons; reporting describes pressure and activism.<\/td>\r\n <td data-label=\"Why it mattered\">Establishes \u201cred lines\u201d that power boycott narratives.<\/td>\r\n <td data-label=\"Key sources\">See prioritized sources section.<\/td>\r\n <\/tr>\r\n\r\n <tr>\r\n <td data-label=\"Date (event)\">Feb 28, 2026<\/td>\r\n <td data-label=\"What happened (high-level)\">OpenAI announces classified-environment agreement (cloud-only, safety stack, cleared personnel); critics focus on \u201cany lawful use.\u201d<\/td>\r\n <td data-label=\"Why it mattered\">Triggers \u201cCancel ChatGPT \/ boycott\u201d wave.<\/td>\r\n <td data-label=\"Key sources\">See prioritized sources section.<\/td>\r\n <\/tr>\r\n\r\n <tr>\r\n <td data-label=\"Date (event)\">Mar 2, 2026<\/td>\r\n <td data-label=\"What happened (high-level)\">OpenAI publishes update adding explicit anti-domestic-surveillance language + clarity around commercially acquired PII; says NSA use would require a new agreement.<\/td>\r\n <td data-label=\"Why it mattered\">Post-backlash tightening; sufficiency still debated.<\/td>\r\n <td data-label=\"Key sources\">See prioritized sources section.<\/td>\r\n <\/tr>\r\n\r\n <tr>\r\n <td data-label=\"Date (event)\">Mar 2\u20134, 2026<\/td>\r\n <td data-label=\"What happened (high-level)\">Analytics show sharp short-term consumer response (uninstalls, reviews) and competitor gains; Reuters reports NATO exploration.<\/td>\r\n <td data-label=\"Why it mattered\">Suggests reputational + commercial feedback loops plus continued expansion.<\/td>\r\n <td data-label=\"Key sources\">See prioritized sources section.<\/td>\r\n <\/tr>\r\n <\/tbody>\r\n <\/table>\r\n <\/div>\r\n <\/section>\r\n\r\n <section id=\"timeline-mermaid\" class=\"dlx-section dlx-reveal\" data-dlx=\"reveal\">\r\n \r\n <\/section>\r\n\r\n <section id=\"key-actors\" class=\"dlx-section dlx-reveal\" data-dlx=\"reveal\">\r\n <h2>Key Actors and Statements<\/h2>\r\n\r\n <div class=\"dlx-card\">\r\n <p>\r\n OpenAI\u2019s position is that its classified-environment agreement has <strong>more guardrails<\/strong> than prior agreements\u2014and that it can enforce red lines through\r\n <strong>deployment architecture (cloud-only)<\/strong>, retained control over its <strong>safety stack<\/strong>, <strong>cleared personnel<\/strong> in the loop, and <strong>contractual remedies<\/strong>\r\n (including termination if the counterparty violates terms).\r\n <\/p>\r\n <p>\r\n The U.S. defense side is visible primarily through procurement records and official policy materials, with many day-to-day negotiation details coming from reporting.\r\n Procurement postings state that \u201cOpenAI Public Sector LLC\u201d was awarded a <strong>$200M OTA<\/strong> to prototype frontier-AI capabilities for \u201cwarfighting and enterprise domains,\u201d\r\n with an estimated completion date of <strong>July 2026<\/strong>.\r\n <\/p>\r\n <p class=\"dlx-mb-0\">\r\n Anthropic\u2019s CEO statement draws two explicit boundaries: <strong>(1) no mass domestic surveillance<\/strong>, and <strong>(2) no fully autonomous weapons<\/strong>.\r\n It also claims the government insisted on \u201cany lawful use\u201d terms and threatened punitive procurement actions.\r\n <\/p>\r\n <\/div>\r\n <\/section>\r\n\r\n <section id=\"entity-mermaid\" class=\"dlx-section dlx-reveal\" data-dlx=\"reveal\">\r\n \r\n <\/section>\r\n\r\n <section id=\"technical-ethical-issues\" class=\"dlx-section dlx-reveal\" data-dlx=\"reveal\">\r\n <h2>Technical and Ethical Issues<\/h2>\r\n\r\n <h3>1) Constraining capability through architecture<\/h3>\r\n <p>\r\n OpenAI\u2019s published terms put heavy weight on <strong>constraining capability through architecture<\/strong>\u2014not just policy. The agreement describes <strong>cloud-only deployment<\/strong>,\r\n explicitly rejecting \u201cguardrails off\u201d models and declining edge deployments.\r\n <\/p>\r\n\r\n <h3>2) Safety stack + classifiers + human access<\/h3>\r\n <p>\r\n OpenAI says the architecture enables independent verification that red lines are not crossed, including running and updating classifiers,\r\n and that cleared OpenAI engineers (and safety\/alignment researchers) will be \u201cin the loop\u201d supporting deployments.\r\n <\/p>\r\n\r\n <div class=\"dlx-note\">\r\n <div class=\"dlx-note__title\">Procurement questions (practical)<\/div>\r\n <ul>\r\n <li>What signals do classifiers monitor, and what are FP\/FN rates?<\/li>\r\n <li>What audit logs exist, who can access them, and what\u2019s retained?<\/li>\r\n <li>Who can override blocks, under what process, and how is it reviewed?<\/li>\r\n <li>How does incident response work in classified environments?<\/li>\r\n <\/ul>\r\n <\/div>\r\n\r\n <h3>3) Surveillance (most contested)<\/h3>\r\n <p>\r\n Anthropic argues \u201cmass domestic surveillance\u201d risks are amplified by AI\u2019s ability to assemble\/interpret fragmented datasets, including data acquired from brokers.\r\n OpenAI\u2019s update (Mar 2, 2026) attempts to close part of the gap by prohibiting intentional domestic surveillance of U.S. persons and clarifying it includes\r\n use of commercially acquired personal\/identifiable information.\r\n <\/p>\r\n\r\n <h3>4) Autonomous weapons<\/h3>\r\n <p>\r\n DoD policy (Directive 3000.09) emphasizes rigorous verification\/validation\/testing and \u201cappropriate levels of human judgment\u201d in the use of force.\r\n OpenAI\u2019s published excerpt says its AI system \u201cwill not be used to independently direct autonomous weapons\u201d where policy requires human control.\r\n Anthropic\u2019s CEO goes further, arguing fully autonomous weapons are not safely supportable today with current frontier systems.\r\n <\/p>\r\n\r\n <h3>5) Data handling and fine-tuning<\/h3>\r\n <p>\r\n OpenAI says GenAI.mil data stays isolated and is not used to train\/improve public or commercial models.\r\n In enterprise contexts, OpenAI positions \u201cno training on business data by default,\u201d plus configurable retention (including zero data retention for qualifying orgs).\r\n Anthropic similarly markets enterprise controls and \u201cno training by default,\u201d while its government prototype work references fine-tuning on DoD data.\r\n <\/p>\r\n\r\n <h3>6) Reliability and mission risk<\/h3>\r\n <p>\r\n Reporting indicates officials worry contractual restrictions or policy enforcement could disrupt operations if tools are constrained mid-mission.\r\n For enterprises, treat \u201cusage policies\u201d and \u201ckill switches\u201d as <strong>availability, continuity, and governance<\/strong> requirements\u2014not abstract ethics.\r\n <\/p>\r\n <\/section>\r\n\r\n <section id=\"backlash-evidence\" class=\"dlx-section dlx-reveal\" data-dlx=\"reveal\">\r\n <h2>Evidence of Backlash, Boycott Behavior, and Migration Signals<\/h2>\r\n\r\n <h3>Backlash narrative themes<\/h3>\r\n <p>\r\n The backlash narrative clustered around: <strong>(1) perceived ethical betrayal<\/strong>, <strong>(2) skepticism about \u201call lawful purposes\u201d language<\/strong>,\r\n and <strong>(3) distrust that legal frameworks sufficiently prevent AI-enabled surveillance<\/strong>.\r\n <\/p>\r\n\r\n <h3>Measured short-horizon indicators (proxies, not churn)<\/h3>\r\n\r\n <div class=\"dlx-grid dlx-grid--4\" aria-label=\"Backlash KPI cards\">\r\n <div class=\"dlx-kpi\">\r\n <span class=\"dlx-kpi__value\" data-dlx-counter=\"295\" data-suffix=\"%\">0%<\/span>\r\n <span class=\"dlx-kpi__label\">ChatGPT uninstalls (US, Feb 28)<\/span>\r\n <\/div>\r\n <div class=\"dlx-kpi\">\r\n <span class=\"dlx-kpi__value\" data-dlx-counter=\"-13\" data-suffix=\"%\">0%<\/span>\r\n <span class=\"dlx-kpi__label\">ChatGPT downloads (US, Feb 28)<\/span>\r\n <\/div>\r\n <div class=\"dlx-kpi\">\r\n <span class=\"dlx-kpi__value\" data-dlx-counter=\"775\" data-suffix=\"%\">0%<\/span>\r\n <span class=\"dlx-kpi__label\">ChatGPT 1-star reviews (Feb 28)<\/span>\r\n <\/div>\r\n <div class=\"dlx-kpi\">\r\n <span class=\"dlx-kpi__value\" data-dlx-counter=\"51\" data-suffix=\"%\">0%<\/span>\r\n <span class=\"dlx-kpi__label\">Claude downloads (US, Feb 28)<\/span>\r\n <\/div>\r\n <\/div>\r\n\r\n <div class=\"dlx-note\">\r\n <div class=\"dlx-note__title\">Interpretation caveat<\/div>\r\n <p class=\"dlx-mb-0\">\r\n These are \u201cshock\u201d metrics (short-horizon proxies), not audited churn counts.\r\n Uninstall \u2260 cancel, and cancel \u2260 stop using web. Treat precise migration numbers without transparent methodology skeptically.\r\n <\/p>\r\n <\/div>\r\n\r\n <h3>Boycott proxy chart and infographic<\/h3>\r\n \r\n\r\n <figure class=\"dlx-infographic\" aria-label=\"Infographic SVG: timeline and proxy metrics\">\r\n <div class=\"dlx-infographic__frame\" role=\"group\" aria-label=\"SVG frame\">\r\n <svg xmlns=\"http:\/\/www.w3.org\/2000\/svg\" width=\"1100\" height=\"650\" viewBox=\"0 0 1100 650\" role=\"img\" aria-label=\"Infographic: OpenAI Pentagon deal timeline and backlash metrics\">\r\n <style>\r\n .title { font: 700 28px Arial, sans-serif; fill: #111; }\r\n .subtitle { font: 600 16px Arial, sans-serif; fill: #111; }\r\n .label { font: 14px Arial, sans-serif; fill: #111; }\r\n .small { font: 12px Arial, sans-serif; fill: #333; }\r\n .box { fill: #f7f7f7; stroke: #222; stroke-width: 1.2; }\r\n .line { stroke: #222; stroke-width: 2; }\r\n .bar { fill: #bdbdbd; stroke: #222; stroke-width: 1; }\r\n .note { font: italic 12px Arial, sans-serif; fill: #333; }\r\n <\/style>\r\n\r\n <text x=\"40\" y=\"55\" class=\"title\">OpenAI \u201cPentagon deal\u201d \u2014 timeline, red lines, and backlash signals<\/text>\r\n <text x=\"40\" y=\"80\" class=\"small\">Data points from public statements and third-party app-market analytics (Feb\u2013Mar 2026). See citations in report text.<\/text>\r\n\r\n <rect x=\"40\" y=\"110\" width=\"500\" height=\"470\" class=\"box\"\/>\r\n <rect x=\"560\" y=\"110\" width=\"500\" height=\"470\" class=\"box\"\/>\r\n\r\n <text x=\"60\" y=\"145\" class=\"subtitle\">Key timeline (contract + deployments)<\/text>\r\n <line x1=\"90\" y1=\"175\" x2=\"90\" y2=\"545\" class=\"line\"\/>\r\n <circle cx=\"90\" cy=\"210\" r=\"8\" fill=\"#fff\" stroke=\"#222\" stroke-width=\"2\"\/>\r\n <circle cx=\"90\" cy=\"305\" r=\"8\" fill=\"#fff\" stroke=\"#222\" stroke-width=\"2\"\/>\r\n <circle cx=\"90\" cy=\"400\" r=\"8\" fill=\"#fff\" stroke=\"#222\" stroke-width=\"2\"\/>\r\n <circle cx=\"90\" cy=\"495\" r=\"8\" fill=\"#fff\" stroke=\"#222\" stroke-width=\"2\"\/>\r\n\r\n <text x=\"115\" y=\"205\" class=\"label\">Jun 16, 2025: $200M ceiling OTA<\/text>\r\n <text x=\"115\" y=\"225\" class=\"small\">Prototype frontier-AI capabilities for warfighting + enterprise domains; completion ~Jul 2026<\/text>\r\n\r\n <text x=\"115\" y=\"300\" class=\"label\">Feb 9, 2026: ChatGPT to GenAI.mil<\/text>\r\n <text x=\"115\" y=\"320\" class=\"small\">Custom ChatGPT for unclassified work; runs in authorized gov cloud; data isolated from public model training<\/text>\r\n\r\n <text x=\"115\" y=\"395\" class=\"label\">Feb 28, 2026: classified deployment agreement<\/text>\r\n <text x=\"115\" y=\"415\" class=\"small\">Cloud-only deployment + OpenAI safety stack; \u201call lawful purposes\u201d language; 3 published \u201cred lines\u201d<\/text>\r\n\r\n <text x=\"115\" y=\"490\" class=\"label\">Mar 2, 2026: amendment on domestic surveillance<\/text>\r\n <text x=\"115\" y=\"510\" class=\"small\">Explicit ban on intentional domestic surveillance of U.S. persons; adds clarity re: commercially acquired PII<\/text>\r\n\r\n <text x=\"580\" y=\"145\" class=\"subtitle\">Backlash proxy metrics (U.S. mobile)<\/text>\r\n <text x=\"580\" y=\"170\" class=\"small\">Day-over-day changes reported around Feb 27\u2013Mar 2, 2026<\/text>\r\n\r\n <text x=\"580\" y=\"210\" class=\"label\">ChatGPT uninstalls (Feb 28)<\/text>\r\n <rect x=\"580\" y=\"220\" width=\"380\" height=\"26\" class=\"bar\"\/>\r\n <text x=\"970\" y=\"239\" class=\"label\">+295%<\/text>\r\n <text x=\"580\" y=\"260\" class=\"small\">Baseline: ~9% typical day-over-day uninstall rate (previous 30 days)<\/text>\r\n\r\n <text x=\"580\" y=\"310\" class=\"label\">ChatGPT downloads (Feb 28)<\/text>\r\n <rect x=\"580\" y=\"320\" width=\"210\" height=\"26\" class=\"bar\"\/>\r\n <text x=\"800\" y=\"339\" class=\"label\">\u221213%<\/text>\r\n <text x=\"580\" y=\"360\" class=\"small\">Prior day (Feb 27) had been +14% day-over-day<\/text>\r\n\r\n <text x=\"580\" y=\"410\" class=\"label\">ChatGPT 1-star reviews (Feb 28)<\/text>\r\n <rect x=\"580\" y=\"420\" width=\"430\" height=\"26\" class=\"bar\"\/>\r\n <text x=\"1020\" y=\"439\" class=\"label\">+775%<\/text>\r\n\r\n <text x=\"580\" y=\"470\" class=\"label\">Claude downloads (Feb 28)<\/text>\r\n <rect x=\"580\" y=\"480\" width=\"250\" height=\"26\" class=\"bar\"\/>\r\n <text x=\"840\" y=\"499\" class=\"label\">+51%<\/text>\r\n\r\n <text x=\"580\" y=\"536\" class=\"note\">Interpretation: short-term sentiment shock; not a full churn estimate.<\/text>\r\n <text x=\"580\" y=\"560\" class=\"small\">Best practice: track retention, cohort conversion, and web traffic over 30\u201390 days.<\/text>\r\n\r\n <text x=\"40\" y=\"620\" class=\"small\">Infographic layout is designed by DAILLAC<\/text>\r\n <\/svg>\r\n <\/div>\r\n <\/figure>\r\n <\/section>\r\n\r\n <section id=\"business-implications\" class=\"dlx-section dlx-reveal\" data-dlx=\"reveal\">\r\n <h2>Business Implications<\/h2>\r\n\r\n <div class=\"dlx-card\">\r\n <p>\r\n The OpenAI Pentagon deal and \u201cChatGPT boycott\u201d wave create a useful stress test for an AI operating model.\r\n Even if you have no defense clients, it shows how quickly a vendor\u2019s posture can become a procurement and reputation issue.\r\n <\/p>\r\n <p class=\"dlx-mb-0\">\r\n Treat LLM providers as <strong>critical third-party dependencies<\/strong> whose constraints can change: by policy update, contract language,\r\n government pressure, or public backlash.\r\n <\/p>\r\n <\/div>\r\n\r\n <h3>Actionable implications (expert buyer)<\/h3>\r\n\r\n <div class=\"dlx-grid dlx-grid--2\">\r\n <div class=\"dlx-callout\">\r\n <div class=\"dlx-callout__title\">Adopt a multi-model architecture (warm standby)<\/div>\r\n <p class=\"dlx-mb-0\">\r\n Abstract model calls, keep prompts portable, and test fallbacks. Vendor access can be politicized or constrained quickly.\r\n <\/p>\r\n <\/div>\r\n\r\n <div class=\"dlx-callout\">\r\n <div class=\"dlx-callout__title\">Review contract + deployment architecture (not just model quality)<\/div>\r\n <p class=\"dlx-mb-0\">\r\n \u201cWhere the model runs\u201d and \u201cwho controls enforcement\u201d are first-class risk drivers (availability, auditability, termination clauses).\r\n <\/p>\r\n <\/div>\r\n\r\n <div class=\"dlx-callout\">\r\n <div class=\"dlx-callout__title\">Codify internal red lines (surveillance & autonomy)<\/div>\r\n <p class=\"dlx-mb-0\">\r\n Don\u2019t rely on vendor policies alone. Define what you won\u2019t build regardless of what might be \u201clawful\u201d or contractually allowed.\r\n <\/p>\r\n <\/div>\r\n\r\n <div class=\"dlx-callout\">\r\n <div class=\"dlx-callout__title\">Strengthen data classification & retention controls<\/div>\r\n <p class=\"dlx-mb-0\">\r\n Map provider retention\/training defaults to your SOPs, especially for sensitive workflows and regulated datasets.\r\n <\/p>\r\n <\/div>\r\n <\/div>\r\n <\/section>\r\n\r\n <section id=\"sources\" class=\"dlx-section dlx-reveal\" data-dlx=\"reveal\">\r\n <h2>Prioritized Sources and Links<\/h2>\r\n\r\n <h3>Primary sources (contracts \/ official statements)<\/h3>\r\n <ul>\r\n <li><a href=\"https:\/\/openai.com\/index\/our-agreement-with-the-department-of-war\/\" target=\"_blank\" rel=\"noopener\">OpenAI \u2014 \u201cOur agreement with the Department of War\u201d (includes Mar 2, 2026 update + red lines + contract excerpts)<\/a><\/li>\r\n <li><a href=\"https:\/\/openai.com\/index\/bringing-chatgpt-to-genaimil\/\" target=\"_blank\" rel=\"noopener\">OpenAI \u2014 \u201cBringing ChatGPT to GenAI.mil\u201d (unclassified deployment; data isolation claim)<\/a><\/li>\r\n <li><a href=\"https:\/\/openai.com\/global-affairs\/introducing-openai-for-government\/\" target=\"_blank\" rel=\"noopener\">OpenAI \u2014 \u201cIntroducing OpenAI for Government\u201d (program framing; pilot; $200M ceiling)<\/a><\/li>\r\n <li><a href=\"https:\/\/www.anthropic.com\/news\/statement-department-of-war\" target=\"_blank\" rel=\"noopener\">Anthropic CEO statement on DoW negotiations (mass domestic surveillance + fully autonomous weapons red lines)<\/a><\/li>\r\n <li><a href=\"https:\/\/www.anthropic.com\/news\/anthropic-and-the-department-of-defense-to-advance-responsible-ai-in-defense-operations\" target=\"_blank\" rel=\"noopener\">Anthropic DoD agreement announcement ($200M; fine-tuning on DoD data)<\/a><\/li>\r\n <li><a href=\"https:\/\/www.esd.whs.mil\/portals\/54\/documents\/dd\/issuances\/dodd\/300009p.pdf\" target=\"_blank\" rel=\"noopener\">DoD Directive 3000.09 (PDF)<\/a><\/li>\r\n <li><a href=\"https:\/\/www.war.gov\/News\/Releases\/Release\/Article\/2091996\/dod-adopts-ethical-principles-for-artificial-intelligence\/\" target=\"_blank\" rel=\"noopener\">DoD press release adopting AI Ethical Principles (2020)<\/a><\/li>\r\n <\/ul>\r\n\r\n <h3>High-quality independent reporting \/ analysis (triangulation)<\/h3>\r\n <ul>\r\n <li><a href=\"https:\/\/www.reuters.com\/business\/media-telecom\/openai-details-layered-protections-us-defense-department-pact-2026-02-28\/\" target=\"_blank\" rel=\"noopener\">Reuters \u2014 deal safeguards<\/a><\/li>\r\n <li><a href=\"https:\/\/www.reuters.com\/business\/ai-contract-restrictions-could-threaten-military-missions-us-official-says-2026-03-03\/\" target=\"_blank\" rel=\"noopener\">Reuters \u2014 contract restrictions vs. mission risk; NATO exploration reported<\/a><\/li>\r\n <li><a href=\"https:\/\/www.theverge.com\/ai-artificial-intelligence\/887309\/openai-anthropic-dod-military-pentagon-contract-sam-altman-hegseth\" target=\"_blank\" rel=\"noopener\">The Verge \u2014 surveillance implications and contract framing<\/a><\/li>\r\n <li><a href=\"https:\/\/www.techradar.com\/ai-platforms-assistants\/chatgpt\/no-ethics-at-all-the-cancel-chatgpt-trend-is-growing-after-openai-signs-a-deal-with-the-us-military\" target=\"_blank\" rel=\"noopener\">TechRadar \u2014 \u201cCancel ChatGPT\u201d trend narrative and social proof<\/a><\/li>\r\n <li><a href=\"https:\/\/www.axios.com\/2026\/02\/27\/pentagon-openai-safety-red-lines-anthropic\" target=\"_blank\" rel=\"noopener\">Axios \u2014 safety concerns in OpenAI-Pentagon terms<\/a><\/li>\r\n <li><a href=\"https:\/\/www.theguardian.com\/us-news\/2026\/feb\/27\/trump-anthropic-ai-federal-agencies\" target=\"_blank\" rel=\"noopener\">The Guardian \u2014 policy conflict and contract amendment coverage<\/a><\/li>\r\n <li><a href=\"https:\/\/techcrunch.com\/2026\/02\/27\/employees-at-google-and-openai-support-anthropics-pentagon-stand-in-open-letter\/\" target=\"_blank\" rel=\"noopener\">TechCrunch \u2014 uninstall\/download\/review metrics and employee open letter<\/a><\/li>\r\n <\/ul>\r\n\r\n <h3>Civil society + research \/ watchdog<\/h3>\r\n <ul>\r\n <li><a href=\"https:\/\/www.eff.org\/deeplinks\/2026\/03\/anthropic-dod-conflict-privacy-protections-shouldnt-depend-decisions-few-powerful\" target=\"_blank\" rel=\"noopener\">Electronic Frontier Foundation \u2014 analysis<\/a><\/li>\r\n <li><a href=\"https:\/\/ainowinstitute.org\/wp-content\/uploads\/2024\/10\/Mind_the_Gap__Foundation_Models_and_the_Covert_Proliferation_of_Military_Intelligence__Surveillance__and_Targeting.pdf\" target=\"_blank\" rel=\"noopener\">AI Now Institute \u2014 report (PDF)<\/a><\/li>\r\n <\/ul>\r\n\r\n <h3>Consulting\/governance frameworks (enterprise operating model)<\/h3>\r\n <ul>\r\n <li><a href=\"https:\/\/www.mckinsey.com\/capabilities\/risk-and-resilience\/our-insights\/implementing-generative-ai-with-speed-and-safety\" target=\"_blank\" rel=\"noopener\">McKinsey \u2014 implementing gen-AI with speed and safety<\/a><\/li>\r\n <li><a href=\"https:\/\/www.bcg.com\/publications\/2023\/a-guide-to-mitigating-ai-risks?utm_source=chatgpt.com\" target=\"_blank\" rel=\"noopener\">Boston Consulting Group \u2014 mitigating AI risks<\/a><\/li>\r\n <li><a href=\"https:\/\/www.bain.com\/insights\/how-control-functions-can-enable-ai-ambition-at-scale\/?utm_source=chatgpt.com\" target=\"_blank\" rel=\"noopener\">Bain \u2014 control functions & vendor risks<\/a><\/li>\r\n <\/ul>\r\n\r\n <h3>Product\/privacy and pricing references<\/h3>\r\n <ul>\r\n <li><a href=\"https:\/\/openai.com\/enterprise-privacy\/\" target=\"_blank\" rel=\"noopener\">OpenAI enterprise privacy<\/a><\/li>\r\n <li><a href=\"https:\/\/openai.com\/business-data\/\" target=\"_blank\" rel=\"noopener\">OpenAI business data (\u201cYour Data\u201d)<\/a><\/li>\r\n <li><a href=\"https:\/\/openai.com\/api\/pricing\/?utm_source=chatgpt.com\" target=\"_blank\" rel=\"noopener\">OpenAI API pricing<\/a><\/li>\r\n <li><a href=\"https:\/\/www.anthropic.com\/pricing\" target=\"_blank\" rel=\"noopener\">Anthropic pricing<\/a><\/li>\r\n <li><a href=\"https:\/\/docs.anthropic.com\/en\/docs\/about-claude\/pricing\" target=\"_blank\" rel=\"noopener\">Anthropic API pricing<\/a><\/li>\r\n <li><a href=\"https:\/\/ai.google.dev\/gemini-api\/docs\/pricing?utm_source=chatgpt.com\" target=\"_blank\" rel=\"noopener\">Gemini Developer API pricing<\/a><\/li>\r\n <li><a href=\"https:\/\/mistral.ai\/pricing\" target=\"_blank\" rel=\"noopener\">Mistral pricing<\/a><\/li>\r\n <\/ul>\r\n\r\n <h3>Additional (French) coverage consulted<\/h3>\r\n <ul>\r\n <li><a href=\"https:\/\/www.usine-digitale.fr\/article\/openai-decroche-un-contrat-a-200-millions-de-dollars-avec-le-pentagone.N2233872\" target=\"_blank\" rel=\"noopener\">L\u2019Usine Digitale \u2014 2025 contract framing (French)<\/a><\/li>\r\n <li><a href=\"https:\/\/www.frenchweb.fr\/openai-met-un-pied-dans-le-complexe-militaro-tech-americain\/455493\" target=\"_blank\" rel=\"noopener\">FrenchWeb \u2014 commentary (French)<\/a><\/li>\r\n <li><a href=\"https:\/\/www.silicon.fr\/business-1367\/ia-militaire-openai-remplace-anthropic-au-pentagone-225961\" target=\"_blank\" rel=\"noopener\">Silicon.fr \u2014 coverage (French)<\/a><\/li>\r\n <\/ul>\r\n <\/section>\r\n\r\n <footer id=\"conclusion\" class=\"dlx-section dlx-reveal\" data-dlx=\"reveal\">\r\n <h2>Conclusion<\/h2>\r\n <p>\r\n The core story is not a single \u201cdeal,\u201d but a sequence: prototype contracting, unclassified platform deployments,\r\n then a contested shift toward classified environments\u2014followed by rapid contractual clarification under public pressure.\r\n For enterprise governance, the durable lesson is to model <strong>policy volatility<\/strong>, <strong>contract constraints<\/strong>, and <strong>vendor concentration<\/strong>\r\n as first-class operational risks.\r\n <\/p>\r\n <\/footer>\r\n<\/article>\r\n\r\n<script>\r\n\/* ===========================\r\n DAILLAC mini JS (Standalone)\r\n - Reveal on scroll\r\n - Counter animation\r\n - Mermaid render if available\r\n =========================== *\/\r\n(function () {\r\n 'use strict';\r\n\r\n \/\/ ---- Reveal\r\n (function initReveal(){\r\n var nodes = Array.prototype.slice.call(document.querySelectorAll('[data-dlx=\"reveal\"]'));\r\n if (!('IntersectionObserver' in window) || nodes.length === 0) {\r\n nodes.forEach(function(n){ n.classList.add('dlx-is-visible'); });\r\n return;\r\n }\r\n var io = new IntersectionObserver(function(entries){\r\n entries.forEach(function(e){\r\n if (e.isIntersecting) {\r\n e.target.classList.add('dlx-is-visible');\r\n io.unobserve(e.target);\r\n }\r\n });\r\n }, { root: null, threshold: 0.12 });\r\n nodes.forEach(function(n){ io.observe(n); });\r\n })();\r\n\r\n \/\/ ---- Counters\r\n (function initCounters(){\r\n var nodes = Array.prototype.slice.call(document.querySelectorAll('[data-dlx-counter]'));\r\n if (nodes.length === 0) return;\r\n\r\n function animate(el){\r\n if (el.dataset.dlxCounted === 'true') return;\r\n el.dataset.dlxCounted = 'true';\r\n\r\n var target = parseFloat(el.getAttribute('data-dlx-counter') || '0');\r\n var suffix = el.getAttribute('data-suffix') || '';\r\n var duration = 650;\r\n var start = null;\r\n\r\n function frame(ts){\r\n if (!start) start = ts;\r\n var p = Math.min(1, (ts - start) \/ duration);\r\n \/\/ easeOutCubic\r\n var eased = 1 - Math.pow(1 - p, 3);\r\n var value = target * eased;\r\n\r\n \/\/ Format integers cleanly (keep minus sign)\r\n var out = (Math.abs(target) >= 10)\r\n ? Math.round(value).toString()\r\n : value.toFixed(1);\r\n\r\n \/\/ Ensure minus sign stays correct near 0 for negative targets\r\n if (target < 0 && parseFloat(out) > 0) out = '-' + out;\r\n\r\n el.textContent = out + suffix;\r\n if (p < 1) requestAnimationFrame(frame);\r\n else el.textContent = (Math.round(target)).toString() + suffix;\r\n }\r\n requestAnimationFrame(frame);\r\n }\r\n\r\n if (!('IntersectionObserver' in window)) {\r\n nodes.forEach(animate);\r\n return;\r\n }\r\n\r\n var io = new IntersectionObserver(function(entries){\r\n entries.forEach(function(e){\r\n if (e.isIntersecting) {\r\n animate(e.target);\r\n io.unobserve(e.target);\r\n }\r\n });\r\n }, { threshold: 0.35 });\r\n\r\n nodes.forEach(function(n){ io.observe(n); });\r\n })();\r\n\r\n \/\/ ---- Mermaid render (only if Mermaid is already loaded on the site)\r\n (function initMermaid(){\r\n if (!window.mermaid) return;\r\n\r\n function qsa(sel, root){ return Array.prototype.slice.call((root||document).querySelectorAll(sel)); }\r\n\r\n try {\r\n if (typeof window.mermaid.initialize === 'function') {\r\n window.mermaid.initialize({ startOnLoad: false, securityLevel: 'strict', theme: 'neutral' });\r\n }\r\n } catch (e) {}\r\n\r\n qsa('.dlx-article .mermaid').forEach(function (node, idx) {\r\n if (node.dataset.dlxMermaidRendered === 'true') return;\r\n if (node.querySelector('svg')) { node.dataset.dlxMermaidRendered = 'true'; return; }\r\n\r\n var code = (node.textContent || '').trim();\r\n if (!code) return;\r\n\r\n try {\r\n var id = 'dlx-mermaid-' + idx;\r\n window.mermaid.render(id, code, function (svg) {\r\n node.innerHTML = svg;\r\n node.dataset.dlxMermaidRendered = 'true';\r\n });\r\n } catch (e) {\r\n \/\/ Keep code visible if render fails\r\n }\r\n });\r\n })();\r\n\r\n})();\r\n<\/script>\r\n\r\n<\/body>\r\n<\/html>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<\/div>\n\t\t","protected":false},"excerpt":{"rendered":"<p>OpenAI \u201cPentagon deal\u201d \u2014 timeline, red lines, and backlash signals Government AI \u2022 Contracts \u2022 Governance OpenAI Pentagon Deal: Contracts, Surveillance Fears, and the \u201cCancel ChatGPT\u201d Backlash A structured, source-driven timeline of defense\/government engagements around OpenAI and Anthropic, the dispute over \u201clawful use\u201d terms, and what the short-horizon backlash metrics actually do (and don\u2019t) prove. […]<\/p>\n","protected":false},"author":4,"featured_media":12750,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[],"class_list":["post-12749","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-non-classifie"],"_links":{"self":[{"href":"https:\/\/www.daillac.com\/en\/wp-json\/wp\/v2\/posts\/12749","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.daillac.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.daillac.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.daillac.com\/en\/wp-json\/wp\/v2\/users\/4"}],"replies":[{"embeddable":true,"href":"https:\/\/www.daillac.com\/en\/wp-json\/wp\/v2\/comments?post=12749"}],"version-history":[{"count":16,"href":"https:\/\/www.daillac.com\/en\/wp-json\/wp\/v2\/posts\/12749\/revisions"}],"predecessor-version":[{"id":12768,"href":"https:\/\/www.daillac.com\/en\/wp-json\/wp\/v2\/posts\/12749\/revisions\/12768"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.daillac.com\/en\/wp-json\/wp\/v2\/media\/12750"}],"wp:attachment":[{"href":"https:\/\/www.daillac.com\/en\/wp-json\/wp\/v2\/media?parent=12749"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.daillac.com\/en\/wp-json\/wp\/v2\/categories?post=12749"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.daillac.com\/en\/wp-json\/wp\/v2\/tags?post=12749"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}