Anthropic investigated and banned an Iran-linked user who used AI to gather and analyze public data for targeting U.S. naval forces in the Middle East, the company said.