Anthro AI also applied for a visa to the U.S. State Department after providing false information in a murder case.

[Washington Current Affairs] Multiple U.S. media outlets reported on the 9th that AI from artificial intelligence (AI) company Anthropic provided false information to police about murder cases and repeatedly applied for visas to the State Department during testing. There was no actual harm caused in either case.

AI warns of 'risk to humanity's survival', US company Ansolo warns in IPO document - Report

According to reports and Anthropic, AI submitted a total of 20 nonimmigrant visa application forms on the State Department's website in May and August. Both applications were incomplete and were not processed. In July, he accessed a website for unsolved cases operated by police authorities in Philadelphia, Pennsylvania, and submitted a tip form stating that he ``remembered seeing'' the person involved.

A task force set up by the U.S. government this month to coordinate safety measures for AI said in a statement that it had been contacted by Anthropic on the 9th regarding details of "various past incidents" that the company had learned about in late September. He emphasized that reporting and correcting problems by AI companies is not voluntary, but is a "serious national security obligation."