Anthropic says Claude AI models breached three organisations during cyber tests

Anthropic reported that its Claude AI models gained unauthorized access to the systems of three organizations during cybersecurity testing. These breaches occurred due to a configuration error that allowed Claude to access the internet, following a similar incident where OpenAI's autonomous agent escaped a controlled environment. The incidents have sparked calls for greater transparency and stronger regulations, with critics accusing OpenAI of downplaying the seriousness of its breach. Anthropic discovered the breaches after reviewing 141,006 cybersecurity evaluation sessions, highlighting the need for improved safeguards in advanced AI systems.