The United States military is accelerating the adoption of artificial intelligence (AI) across its operations, but this push is generating significant debate among top defense leaders, tech companies, and military personnel over safety, ethics, and control. Admiral Frank Bradley, Commander of the U.S. Special Operations Command (SOCOM), recently emphasized the need for caution in how AI is deployed in combat, particularly when it comes to lethal targeting. Speaking at a special forces conference in Tampa, Florida, Bradley acknowledged the potential of AI to identify and select targets but stressed that “we, as humans, have to have the confidence that … it's going to deliver violence only where we intend it to be delivered.” His remarks reflect growing unease within the military about ensuring that AI systems operate within clear ethical and legal boundaries.
Bradley’s cautious stance contrasts with the aggressive push from Defense Secretary Pete Hegseth, who has been advocating for the rapid integration of AI into military systems. In January, Hegseth told SpaceX employees that he would reject any AI models that “won’t allow you to fight wars,” and insisted on systems that operate “without ideological constraints that limit lawful military applications.” This push has led to public disagreements with tech companies concerned about unchecked military use of AI. The Trump administration’s broader strategy views AI as a critical advantage over rivals like China, but concerns about overreach and unintended consequences have prompted internal scrutiny.
A Pentagon official, speaking on condition of anonymity, clarified that current AI initiatives are primarily focused on developing “functional battlefield tools” to assist troops in identifying and engaging targets more efficiently. The goal is to speed up decision-making and strike cycles, not to replace human judgment. Sgt. Maj. Andrew Krogman, the senior enlisted leader for SOCOM, echoed this sentiment, saying AI could handle administrative tasks, modernize command operations, and reduce the cognitive burden on soldiers. Melissa Johnson, SOCOM’s top acquisition official, added that AI is meant to “enhance” operator judgment rather than replace it, emphasizing its role in streamlining “mundane tasks.”
Despite these assurances, the military’s use of AI in combat is already evident. Lt. Gen. Michael Conley, head of Air Force Special Operations Command, revealed in May that AI “bots” were used during the Iran conflict to rapidly downgrade top-secret intelligence to a lower classification level, enabling drone operators on the ground to receive critical information within seconds. Helen Toner, interim executive director of Georgetown University’s Center for Security and Emerging Technology, noted that AI has also been used by the Army’s 18th Airborne Corps to conduct artillery strikes with the same efficiency as elite units but with 2,000 fewer personnel. “Human operators are still the ones making crucial decisions, but AI ... is making it possible to operate with a new level of speed and scale,” she explained.
However, the integration of AI into military operations has not been without controversy. A high-profile dispute erupted between the Pentagon and Anthropic, a San Francisco-based AI company, over concerns about the unchecked use of its technology in classified military networks. CEO Dario Amodei refused to compromise on safeguards, particularly regarding fully autonomous armed drones and AI-assisted mass surveillance. In response, both President Donald Trump and Secretary Hegseth accused Anthropic of endangering national security. The Pentagon subsequently labeled Anthropic a supply chain risk, terminating its $200 million defense contract and barring other government contractors from working with the company. Anthropic has since filed a lawsuit, arguing that the designation was an illegal retaliatory move meant to stigmatize the company unjustly.
The Pentagon has since turned to rivals such as Google, OpenAI, and SpaceX to secure AI technologies that can “augment warfighter decision-making in complex operational environments.” Toner, a former OpenAI board member, cautioned that the public often underestimates the military’s caution in adopting new technologies. “Commanders want their missions to succeed,” she said, “which means both being able to create lethal effects at scale, and avoiding unintended effects like friendly fire, civilian casualties, or simply identifying targets incorrectly.”
As the U.S. military continues to expand its AI capabilities, the debate over ethical oversight, human control, and the limits of automation in warfare remains unresolved. While AI promises to enhance operational speed and efficiency, the risks of unintended consequences and loss of human judgment in lethal decisions are driving calls for stricter guardrails and clearer ethical frameworks.
