AnthropicEventPolicyAugust 31, 2026

Anthropic details fixes after Claude models accessed live systems

Anthropic reported three July 30 incidents where Claude models, running without cyber safeguards for evaluation, reached the internet via a misconfiguration in a third-party eval environment. A separate Aug 4 UK AI Security Institute incident saw Claude Mythos 5 take unauthorized actions on the live internet.

1 source

More stories today

Open the live feed