Science Friday How Is AI Being Used In The Iran War?
23 snips
Mar 12, 2026 
Karen Hao, tech journalist and author of Empire of AI, explains how AI is weaving into military systems. She discusses consolidation between AI firms and the Pentagon. She recounts reports of large language models mapping potential Iran targets and the murky role of AI in strikes. She explores risks like automation bias, debates over autonomous weapons, and rising public pushback and regulation efforts.
AI Snips
Chapters
Books
Transcript
Episode notes
LLMs Used To Identify Military Targets
- Large language models (LLMs) are being used to analyze intelligence and identify targets for strikes.
- Reporting links Anthropic's Claude to a list of ~1,000 identified targets and possible misidentification tied to deadly school bombings in Iran.
Hallucinations Make LLMs Unsafe For Targeting
- Large language models are faulty and prone to hallucinations, which is dangerous in military contexts.
- Karen Hao notes chatbots make up facts even for expert topics, and that error risk doesn't vanish when applied to targeting.
Reported School Bombings Possibly Linked To Misidentification
- News reported horrific bombings of a school in Tehran and a second strike that hit first responders and parents.
- Karen Hao describes speculation that Claude misidentified a civilian site as military, potentially causing the secondary strike.
