Pentagon pushes battlefield AI

The Pentagon’s latest AI push is moving on two tracks at once. On May 1, the Defense Department said it had reached agreements with seven tech companies to use their artificial intelligence tools on classified military networks, naming Google, Microsoft, Amazon Web Services, Nvidia, OpenAI, Reflection and SpaceX. The department said the systems would help “augment warfighter decision-making in complex operational environments.” At the same time, senior military officers and some AI companies have continued to warn that battlefield use raises separate questions about human control, targeting and accountability. That split reflects a broader policy already laid down in Washington. A January 9, 2026 Pentagon AI strategy directed the department to become an “AI-first” warfighting force and said AI-enabled warfare would reshape military affairs over the next decade. A separate Defense Department responsible AI pathway says the military must apply lawful, ethical and accountable standards when designing, testing, procuring, deploying and using AI. (wsls.com) ### Which companies are actually inside this new Pentagon push? The Defense Department said on May 1 that Google, Microsoft, Amazon Web Services, Nvidia, OpenAI, Reflection and SpaceX would provide AI resources for classified environments. IBTimes reported the tools would be integrated into Impact Level 6 and Impact Level 7 networks, the Defense Department’s sensitive secret and top-secret systems. The Pentagon said the goal was to improve data synthesis, situational understanding and battlefield decision support, while avoiding dependence on a single vendor. (media.defense.gov) Anthropic was not included in that list. IBTimes reported the company had been in a dispute with the Pentagon over whether its AI had to be available for any lawful military use, including concerns about domestic surveillance and fully autonomous lethal weapons. The same report said the Pentagon described Anthropic as a supply-chain risk and that the company sued in response. (wsls.com) ### What are military leaders saying about lethal use? Adm. Frank Bradley, the head of U.S. Special Operations Command, told a special forces conference in Tampa that troops “have to be very careful” about AI’s role in “the delivery of lethality.” Bradley said he could see a future in which AI helps determine what targets to hit, but added that “we, as humans, have to have the confidence” that violence is delivered only where intended. (ibtimes.com) Defense Secretary Pete Hegseth has taken a different public line. The Associated Press, in the WSLS report, said Hegseth has argued the Pentagon should be allowed to use AI in any legal way it sees fit. The same report said he told SpaceX employees in January that he would reject AI models “that won’t allow you to fight wars” and wanted systems operating “without ideological constraints that limit lawful military applications.” (wsls.com) ### What does the Pentagon say these systems will do? The Pentagon said the new agreements would help troops make faster decisions in complex operations. The Associated Press reported AI could shorten the time needed to identify and strike targets, while also helping with weapons maintenance and supply lines. Helen Toner of Georgetown University’s Center for Security and Emerging Technology said AI can be useful for summarizing information and scanning surveillance feeds to identify potential targets. (wsls.com) The Pentagon’s own January strategy describes a wider buildout. That document said GenAI.mil would put leading AI models in the hands of the department’s roughly three million civilian and military personnel across classification levels, as part of a broader effort to speed experimentation and fielding. ### Where do the guardrails come from? (wsls.com) The Defense Department’s responsible AI pathway says trust in military AI depends on lawful and ethical behavior in design, testing, procurement, deployment and use. That document presents responsible AI as a way to make policy operational across the department, rather than leaving safeguards as a general principle. (media.defense.gov) Defense cloud and cybersecurity rules also shape the rollout. Defense Department cloud guidance says new secret and top-secret cloud capabilities are to use the Joint Warfighting Cloud Capability contract vehicle, and separate DoD guidance says AI systems need mission-specific security objectives set early in development. The next visible milestone is implementation inside classified networks. (media.defense.gov) The Pentagon has said the seven companies’ tools are headed for Impact Level 6 and 7 environments, while the department’s January AI strategy calls for a series of “pace-setting projects” to accelerate fielding across the force. (ibtimes.com) (dodcio.defense.gov)