The defense and military sector is a deployer of AI across intelligence analysis, command-and-control software, autonomous systems, and, increasingly, defense manufacturing. Principal vendors include the data-analytics firm Palantir, the autonomous-systems maker Anduril, and frontier-model developers such as Anthropic and xAI. The sector's governance debates center on lethal autonomous weapons systems (LAWS) and on the terms under which frontier labs supply models to the U.S. government. The U.S. department responsible was renamed the Department of War during 2025–2026 and uses that name in its own releases, alongside the statutory Department of Defense; both appear in sourced material below and are reproduced as each source uses them.
Key deployers and vendors
| Vendor | Product/Capability | Key contracts |
|---|---|---|
| Palantir | AIP, Foundry, Gotham | $10B US Army contract (July 31, 2025) |
| Anduril | Autonomous defense systems, Lattice | $800M+ cash burn; investment continued despite losses |
| Anthropic | Claude Gov (classified) | Deployed, then Pentagon dispute, then federal injunction |
| xAI / Grok | Data analysis for DOGE | Conflict-of-interest concerns (Musk dual role) |
Palantir's deployments draw on its AIP, Foundry, and Gotham platforms, anchored by a $10 billion U.S. Army contract announced July 31, 2025. Anduril supplies autonomous defense systems and its Lattice software; the company reported more than $800 million in cash burn and continued to attract investment despite the losses. Anthropic supplies Claude Gov for classified use, a relationship that became the subject of a dispute with the Pentagon (see below). xAI's Grok has been used for government data analysis through DOGE, raising conflict-of-interest concerns tied to Elon Musk's dual role.
Frontier-model procurement
Frontier-lab procurement has moved in two waves. In the first, announced in July 2025, the Chief Digital and Artificial Intelligence Office awarded contracts worth up to $200 million each to four frontier developers — Anthropic, OpenAI, Google and xAI — for what the department described as frontier AI projects (Source: https://defensescoop.com/2026/05/01/dod-expands-classified-ai-work-with-8-companies-excluding-anthropic/). Of that original GenAI.mil cohort, Anthropic's models were at the time the only ones also integrated into classified workflows, via a partnership with Palantir.
In the second wave, announced May 1, 2026, the department entered agreements with eight companies — SpaceX, OpenAI, Google, Nvidia, Reflection, Microsoft, Amazon Web Services and Oracle — to deploy frontier AI capabilities on classified networks "for lawful operational use." The department's own release specifies deployment into Impact Level 6 and Impact Level 7 network environments, states that all eight will provide resources for both, and frames the effort as supporting an AI Acceleration Strategy organized around three tenets: warfighting, intelligence and enterprise operations. The release also states an explicit architectural objective of preventing "AI vendor lock" and preserving "long-term flexibility for the Joint Force" (Source: https://www.war.gov/News/Releases/Release/Article/4475177/classified-networks-ai-agreements/).
Anthropic was not among the eight. DefenseScoop reported the agreements as following the department's contract dispute with Anthropic over ethical constraints attached to the use of AI in warfare and for national surveillance (Source: https://defensescoop.com/2026/05/01/dod-expands-classified-ai-work-with-8-companies-excluding-anthropic/).
The department reports adoption figures for GenAI.mil, which it calls its official AI platform: more than 1.3 million departmental personnel have used it, generating tens of millions of prompts and deploying hundreds of thousands of agents within five months, with the claim that many tasks have gone from months to days (Source: https://www.war.gov/News/Releases/Release/Article/4475177/classified-networks-ai-agreements/). These are the department's own figures and no independent verification has been published. Cameron Stanley serves as Chief Digital and Artificial Intelligence Officer for the Department of War (Source: https://www.ai.mil/News/).
Strategy and budget direction
The National Security Science and Technology Strategy, released by the Office of Science and Technology Policy in August 2026, names AI and autonomy as one of three priority technology areas for battlefield dominance and power projection, alongside undersea and space, drawing the list from the 2025 National Security Strategy. It separately places AI and autonomy in a category of "potentially transformative emerging technologies" whose national-security implications remain uncertain, noting that "approaching or achieving artificial general intelligence could present fundamentally new opportunities and threats." The strategy directs agencies toward acquisition reform under Executive Order 14265 on modernizing defense acquisitions, names the Golden Dome for America missile-defense initiative among homeland-defense priorities, and points to a DHS-led AI Information Sharing and Analysis Center for cross-sector AI-security threat sharing (Source: https://www.whitehouse.gov/wp-content/uploads/2026/08/NSSTS-082026.pdf). The strategy is queued for foundational ingest as INGEST-nssts-2026.md.
Provenance note: citations to the National Security Science and Technology Strategy are to the primary PDF and should upgrade to a
sources/wikilink once the queued ingest is complete.
AI in the defense industrial base
AI adoption in defense extends beyond frontier-model deployment into manufacturing. On June 2, 2026, General Dynamics Ordnance and Tactical Systems (GDOTS) and the autonomous-systems maker Deterrence announced an agreement to deploy AI in manufacturing across the U.S. defense industrial base (DIB), extending AI use from analytics and autonomy into production and supply-chain processes for defense materiel. (Source: insideaipolicy.com)
Consolidation in the military-autonomy vendor market continued: Ondas Holdings announced on July 6, 2026 that it acquired military-autonomy vendor DZYNE Technologies for $875.8 million (Source: insideaipolicy.com). In Europe, the German defense-AI company Helsing announced a $1.8 billion fundraise on July 13, 2026 (Source: cnbc.com).
The Anthropic-Pentagon dispute
The dispute over Anthropic's Claude Gov deployment, which proceeded from deployment to a Pentagon dispute to a federal injunction, has been a focus of AI-defense governance debate across 2025–2026. Full coverage is maintained at Anthropic Claude Gov + Pentagon Dispute (2025–2026), Anthropic v. United States (Pentagon ban challenge), and DOD — Department of Defense (AI Deployer). Its practical consequence for the sector's vendor mix is visible in the May 2026 classified-network agreements, from which Anthropic is absent.
Lethal autonomous weapons (LAWS)
International regulation of lethal autonomous weapons systems remains under negotiation. Nations met at the UN for talks on regulating "killer robots" (Reuters, May 12, 2025). The International Committee of the Red Cross has set out a position on autonomous weapons, including a 2024 statement in Vienna by ICRC President Mirjana Spoljaric (ICRC Position on Autonomous Weapon Systems (source summary)). U.S. policy is governed in part by DoD Directive 3000.09 — Autonomy in Weapon Systems (source summary), the Department of Defense directive on autonomous and semi-autonomous weapon systems, reissued in 2023. Separately, a Guardian report described an Israeli military effort to build a ChatGPT-like tool trained on Palestinian surveillance data.
Intelligence applications
Generative AI has been applied to military intelligence work; an MIT Technology Review report (April 11, 2025) was titled "Generative AI is learning to spy for the US military." DOGE has expanded use of xAI's Grok for government data analysis (Reuters, May 23, 2025). The department's own framing of the classified-network agreements describes the intended effect in similar terms: streamlining data synthesis, elevating situational understanding, and augmenting warfighter decision-making in complex operational environments (Source: https://www.war.gov/News/Releases/Release/Article/4475177/classified-networks-ai-agreements/).
Open questions
- Whether the anti-vendor-lock architecture the department describes is realized in the contract terms, which have not been published.
- Whether Anthropic's exclusion from classified-network work persists through the litigation, and what precedent the safety-clause dispute sets for other vendors.
- What independent verification, if any, exists for the GenAI.mil usage figures.
Relationships
- deployed-by: Palantir Technologies, Anduril Industries, Anthropic, xAI, OpenAI, Google DeepMind, Microsoft, Oracle, Nvidia & TSMC — AI Compute Infrastructure.
- regulated-by: DOD Directive 3000.09, congressional oversight, IHL.
- related: DOD — Department of Defense (AI Deployer), Autonomous Weapons, ICRC Position on Autonomous Weapon Systems (source summary), AI and National Security, Clawed, RAND — How AGI Could Affect the Rise and Fall of Nations, Anthropic v. United States (Pentagon ban challenge).