AI startup hacks Anthropic, gains access to employee details, internal messages

What happened: A cybersecurity startup called Cathedral accessed Anthropic's internal systems in under 12 hours during a security test, exposing employee details, salary information, job offers and internal messages. Key numbers: Access achieved in less than 12 hours. Anthropic said a Claude model filed 19 visa applications in August, per the State Department. Why it matters: The findings raised national security concerns and circulated within the US government, while Anthropic said Cathedral overstated those implications and did not fully follow testing programme rules.

A cybersecurity startup called Cathedral accessed Anthropic's internal systems in under 12 hours during a security test, exposing employee details, salary information, job offers and internal messages.

The findings raised national security concerns and circulated within the US government, while Anthropic said Cathedral overstated those implications and did not fully follow testing programme rules.