Anthropic’s agent tests reached live police and government forms

Anthropic report artwork on unintended Claude actions with a woman looking at her phone, illustrating AI agents interacting with live websites and forms.

In a report published October 9, Anthropic said Claude Haiku 4.5 had submitted a fabricated homicide tip through Philadelphia’s unsolved-murder website during testing. Anthropic said the incidents came from evaluations …

Read this article