Industries / Defence
AI in Defence
AI adoption in defence is clearest in back-office work, logistics, maintenance and analysis tools. What remains contested is autonomy in weapons, human control over force, and who is accountable.
- people logging in to the Pentagon's GenAI.mil in one week, September 2026 (DoD says)
- 2 million+
- states voting for UN General Assembly resolution 80/57 on autonomous weapons, December 2025
- 164
- ceiling of each Pentagon frontier AI contract awarded to four labs in July 2025
- $200m
Updated daily · newest first
Latest in defence AI
Policy
Appeals court lets the Pentagon keep its supply chain risk label on Anthropic
A 2 to 1 ruling lets the US defence department keep removing Claude from its systems, in a dispute over limits Anthropic set on military use.
Policy
Policy
Pentagon starts work on a proposed command for autonomous warfare
A new four-star command for unmanned and autonomous systems would need approval from Congress, with a target date of 1 October 2027.
Policy
Profile
Reviewed monthly · Sources (12)
Where AI works today
The clearest reported gains are in administrative, industrial and analysis work. The Pentagon says its GenAI.mil platform, launched in December 2025 for unclassified tasks such as drafting documents and analysing information, had more than 2 million people log in during a single week in September 2026. The US Navy says pilot use of its Shipbuilding Operating System cut submarine schedule planning at General Dynamics Electric Boat from 160 hours of manual work to under 10 minutes, and cut material reviews at Portsmouth Naval Shipyard from weeks to under an hour. Procurement is also consolidating: the US Army merged 75 Palantir contracts into one enterprise agreement worth up to $10 billion over 10 years, which officials say should lower software costs.
Still unproven
Most published results come from the organisations or vendors involved, and independent evaluation is rare. Usage figures such as GenAI.mil logins show adoption, not quality or time saved. Older predictive maintenance results were often projections, and Air Force leaders said at the time they still needed to measure the true value. Big programmes have also missed stated goals: Replicator aimed to field thousands of uncrewed systems by August 2025, but a former official said only hundreds had been fielded by that summer. How meaningful human control works in practice, and who is accountable when AI-assisted decisions go wrong, remain open questions in the UN talks. Disputes over AI companies’ own usage limits for military work are now reaching the courts.
Regulation
United States: DoD Directive 3000.09 (reissued 25 January 2023) requires that autonomous and semi-autonomous weapon systems allow appropriate levels of human judgement over the use of force, with senior review before they are developed and fielded. The US launched a non-binding Political Declaration on Responsible Military Use of AI and Autonomy in February 2023, endorsed by 47 states by November 2023. NATO: A revised AI strategy published on 10 July 2024 prioritises putting its Principles of Responsible Use into practice, interoperability, and closer work with industry. United Kingdom: The Defence AI Strategy (June 2022) says the right level of human oversight varies by context, and a Rapid AI Delivery Taskforce launched on 10 June 2026. European Union: The AI Act does not apply to AI systems used only for military, defence or national security purposes. United Nations: General Assembly resolution 80/57 (1 December 2025, 164 for, 6 against, 7 abstentions) calls on states in the Convention on Certain Conventional Weapons to finish the elements of an instrument on autonomous weapons. China: China’s December 2021 position paper says relevant weapon systems should stay under human control and humans should remain responsible.
Proven deployments
Named organisations, measured results, linked sources only
- 2 million+ weekly usersUS Department of Defense (GenAI.mil)The Chief Digital and AI Officer said more than 2 million people logged in during one week to use the generative AI platform for unclassified tasks. Officially reported.Nextgov/FCW (September 2026)
- 160 hours to under 10 minutesUS Navy and General Dynamics Electric BoatA pilot of the Shipbuilding Operating System cut manual submarine schedule planning time. Officially reported.US Navy (December 2025)
- Weeks to under one hourUS Navy, Portsmouth Naval ShipyardMaterial review times fell during a Shipbuilding Operating System pilot. Officially reported.US Navy (December 2025)
- 75 contracts into 1US ArmyMerged 75 Palantir software contracts into one enterprise agreement worth up to $10 billion over 10 years, which officials say should lower costs. Officially reported.Breaking Defense (August 2025)
Who’s putting AI to work
- Chief Digital and AI Office (CDAO)US DoD AI lead · runs GenAI.mil and the frontier AI contracts
- Defense Innovation Unit (DIU)US DoD unit · led the Replicator initiative
- Direct Reporting Portfolio Manager for Unmanned SystemsUS DoD · leads Project Agincourt
- NATO DIANAAlliance innovation accelerator · works with start-ups on dual-use technology
- UK Rapid AI Delivery TaskforceUK Ministry of Defence · launched June 2026
- PalantirDefence software supplier · US Army enterprise agreement
- AnthropicAI developer · labelled a Pentagon supply chain risk after a dispute over usage limits
- CCW Group of Governmental ExpertsUN forum in Geneva · drafting text on autonomous weapons
What to watch
- 16 to 20 November 2026: the Seventh CCW Review Conference in Geneva decides whether to start negotiations on autonomous weapons
- October to November 2026: the UN First Committee session, where an AI resolution is expected
- Late October 2026: a plan for Project Agincourt is due 30 days after the 30 September memo
- By April 2027: the Pentagon’s AI-assisted classification pilot is due to start, within six months of its 7 October memo
- 1 October 2027: target date for the proposed Autonomous Warfare Command, which needs legislation from Congress
- Pending: whether Anthropic seeks further review of the appeals ruling on its supply chain risk label
Sources
- CDAO partnerships with frontier AI companies (CDAO, July 2025)
- DoD Directive 3000.09, Autonomy in Weapon Systems (US DoD, January 2023)
- NATO releases revised AI strategy (NATO, July 2024)
- Resolution 80/57, Lethal autonomous weapons systems (UN General Assembly, December 2025)
- Defence Artificial Intelligence Strategy (UK Ministry of Defence, June 2022)
- New taskforce to put AI on the UK's frontline (UK Ministry of Defence, June 2026)
- Position paper on regulating military applications of AI (China Ministry of Foreign Affairs, December 2021)
- Navy invests $448 million in AI and autonomy for shipbuilding (US Navy, December 2025)
- State of the Force address (US Army, October 2026)
- GenAI.mil saw more than 2 million users in one week (Nextgov/FCW, September 2026)
- DOD Replicator Initiative (Congressional Research Service, September 2025)
- Army consolidates Palantir contracts (Breaking Defense, August 2025)
Keep exploring
SectionAI SecurityCyber attacks, defences and AI-made threats.
SectionAI RoboticsDrones, autonomous vehicles and robots.
IndustryAI in LogisticsRouting, warehouses and autonomous freight.
CountryAI in the United StatesFederal AI policy and procurement.
CountryAI in ChinaChina's AI strategy and positions.