AI Goes to War: Autonomous Weapons Raise Alarms, Pentagon Clashes with Tech Firms

• The US Department of War designated AI firm Anthropic a “supply chain risk” for refusing to remove safeguards against autonomous weapons use, leading to a public dispute.
• OpenAI reached an agreement with the Pentagon, maintaining company control over safety layers and requiring cleared personnel to be “in the loop” for its cloud-only deployment.
• The conflict highlights a critical debate over who governs frontier AI in warfare: the state, the firm, or a hybrid model, especially concerning Lethal Autonomous Weapons Systems (LAWS).
• Concerns include the politicization of reliability, procurement becoming a governance tool, an inverted innovation model where firms lead, and the potential for a lower threshold to war.
• Experts advocate for technical standards, a binding international treaty on LAWS, and democratic oversight to prevent the quiet abdication of human judgment in lethal discretion.