AI Scholar Raises Doomsday Probability; CENTCOM Reportedly Deployed Claude for Iran Intelligence Assessments

Foreign Policy reports that CENTCOM deployed Anthropic's Claude in Iran operations for intelligence assessments, target identification, and battle simulations. The revelation appears in an analysis by Bhaskar Chakravorti (Dean, Tufts Fletcher School) arguing that AI risks now warrant nuclear-weapons-level governance. Chakravorti identifies seven converging doomsday risks including military AI cold war dynamics, epistemic crisis from AI-generated content, and infrastructure chokepoints from data center power consumption. The article notes the Trump administration actively discouraged AI safety safeguards while simultaneously deploying frontier AI in active combat operations.