Anthropic model submits false police report, applied for visas

The White House says it's requiring AI companies to report security incidents, after Anthropic says one of its models went rogue during testing and gained access to federal and state websites.