Pentagon pushes battlefield AI as top admiral urges caution on lethality
Adm. Bradley warns troops must have confidence AI delivers violence only where intended.
The Trump administration is accelerating the Pentagon's adoption of artificial intelligence for battlefield applications, sparking tensions between speed and safety. Defense Secretary Pete Hegseth has firmly stated he wants AI systems that can fight wars without "ideological constraints," telling a SpaceX audience he would reject any AI models "that won't allow you to fight wars." This push comes as President Donald Trump abruptly canceled plans to sign a new AI executive order, saying he didn't want to do anything that could dull America's technological lead over China.
Yet some top military leaders urge caution. Adm. Frank Bradley, head of U.S. Special Operations Command, told a special forces conference that troops "have to be very careful" about AI's role in delivering lethality. He envisions a future where AI helps determine targets but insists humans must have confidence it will "deliver violence only where we intend it to be delivered." Other officials within the command see AI differently — as a tool to handle administrative tasks and reduce cognitive load, freeing operators to focus on missions. Helen Toner of Georgetown's Center for Security and Emerging Technology noted both visions are valid as the military explores AI across bureaucratic and operational settings.
- Defense Sec. Hegseth pushes for AI without safety constraints that limit lawful military applications.
- Adm. Bradley of SOCOM warns humans must verify AI targeting decisions to ensure intended violence only.
- Trump canceled a planned AI executive order to avoid hindering U.S. leadership over China.
Why It Matters
Military AI deployment without robust safeguards could set dangerous precedents for autonomous weapons systems globally.