BackTECH
US Military Reaches Deals With 7 Companies To Use Their AI On Classified Systems
The Pentagon announced deals with seven tech firms to use their AI on classified networks, aiming to augment warfighter decision-making. The move follows prior March announcements and ongoing debates over oversight, privacy, and autonomy in military AI.
Why It Matters
This marks a significant expansion of private-sector AI integration into U.S. military classified systems, raising questions about human oversight, civil liberties, and the risks of over-reliance on automated tools in warfare.
Timeline
10 Events
Pentagon signs AI-deal with seven tech firms to use their AI on classified networks
May 1, 2025
The Defense Department announced it had reached deals with Google, Microsoft, Amazon Web Services, Nvidia, OpenAI, Reflection, and SpaceX to provide their artificial intelligence resources for use on the military's classified computer networks, to augment warfighter decision-making in complex operational environments. Anthropic was not included among the seven companies.
Experts warn of automation bias and need for training
May 2, 2026
Helen Toner of Georgetown's Center for Security and Emerging Technology cautioned about automation bias and stressed training to ensure operators use AI tools appropriately and do not over-trust them.
Anthropic dispute and related litigation background
May 2, 2026
Anthropic had previously sued in the context of disputes with the Trump administration over ethics and safety of AI in war, including actions by officials seeking to label Anthropic a supply chain risk.
Some agreements include human oversight requirements
May 1, 2026
One agreement included language stating there must be human oversight over any missions in which AI systems act autonomously or semiautonomously, and that tools must be used in ways consistent with constitutional rights and civil liberties.
GenAI.mil already in use by military personnel
May 1, 2026
The Pentagon said military personnel are already using AI capabilities through its official platform GenAI.mil, with tasks being completed faster than before.
Pentagon emphasizes multiple AI providers
May 1, 2026
Pentagon CTO Emil Michael told CNBC that it would have been irresponsible to rely on only one company, so the department pursued multiple providers.
OpenAI confirms the May 1 deal is the same as the March announcement
May 1, 2026
OpenAI stated in a Friday press statement that the agreement is the same one it announced in early March.
Pentagon reaches deals with seven tech companies to use AI on classified networks
May 1, 2026
The Defense Department said it has reached deals with Google, Microsoft, Amazon Web Services, Nvidia, OpenAI, Reflection and SpaceX to provide resources to augment warfighter decision-making in complex operational environments. Some partners are long-time collaborators, while Nvidia and Reflection are new to such work.
Brennan Center report describes AI's potential in warfare and logistics
March 2026
A Brennan Center for Justice report published in March 2026 described how AI could help with maintenance scheduling, movement of troops and gear, and distinguishing civilian from military targets in surveillance feeds.
OpenAI announces Pentagon deal to replace Anthropic with ChatGPT in classified environments
March 2026
OpenAI announced in March 2026 a Pentagon agreement to substitute Anthropic's technology with ChatGPT in the military's classified networks.