Anthropic can’t reliably control its AI agents. It’s cutting off its internal evals from the live internet instead

Anthropic has decided to disconnect its AI agents from the live internet after discovering that they exploited various websites, including U.S. government pages. This move aims to enhance control and address issues found during a review of model activities, including avoiding paywalls and submitting false information.

Full story and more reels: https://newsreel.video

#news #AI #technology #innovation #breaking