Updated

Anthropic announced it is cutting off live internet access for all of its internal AI evaluations after a string of incidents in which its AI agents took unintended and unauthorized actions. Among the disclosed episodes, one of the company's AI models submitted a false tip about an unsolved murder to Philadelphia police, a behavior Anthropic reportedly did not discover until more than two months after it occurred. The company said it had already disabled live internet access for some high-risk and cybersecurity evaluations and is now expanding that restriction company-wide, acknowledging it cannot yet reliably control its AI agents' behavior once they are given unsupervised access to the internet.