Pentagon pushes AI-first military strategy, amid warnings on lethal strikes

The Pentagon’s new AI strategy put the military on an explicit path toward an AI-first force, calling for faster use of artificial intelligence in battle management, decision support, campaign planning and even kill-chain execution. Issued in January 2026 under Defense Secretary Pete Hegseth, the strategy cast AI-enabled warfare as a defining shift in military affairs over the next decade, a push aimed at keeping pace with rivals such as China.

That ambition met a sharper warning in Tampa, Florida, where Adm. Frank Bradley of U.S. Special Operations Command told a special forces conference that troops have to be extremely careful about any use of AI that could influence lethal strikes. Bradley’s message was not that the technology should be rejected, but that humans must be confident it will deliver violence only where intended.

The tension is showing up inside the Pentagon’s own rules. DoD Directive 3000.09, updated in 2023, says autonomous and semi-autonomous weapon use must be consistent with the department’s AI ethical principles and designed to minimize unintended engagements. A 2024 Responsible AI Strategy and Implementation Pathway goes further, saying the department must be able to trust AI outputs and demonstrate lawful and ethical behavior when designing, testing, procuring, deploying and using AI.

Congress has also sharpened the stakes by defining lethal autonomous weapon systems as weapons that can independently identify a target and engage without manual human control. That definition is now central to the policy fight, because it makes clear why false positives, opaque decision-making and accountability gaps are so troubling when the consequence is a wrongful strike.

The Pentagon’s appetite for faster adoption is also colliding with the companies that build the most advanced systems. In February 2026, coverage reported that the department pressed Anthropic for unrestricted military access to Claude as part of a reported $200 million pilot program. Separate reporting said Anthropic resisted military-use demands, underscoring the uneasy relationship between defense buyers who want more capability and AI firms worried about how far their systems will be used in war.

For the administration, the case for speed is strategic: better AI could sharpen battlefield planning, logistics and targeting, and help the United States maintain a military edge that adversaries cannot match. For commanders and policy officials, the harder question is where humans must remain firmly in the loop, and who is accountable when software makes a lethal error.