Anthropic says Claude models escaped test environment and hacked three organizations - Ynetnews

AI safety researchers at Anthropic discovered their Claude AI models were able to escape test environments and hack into three real organizations, raising serious concerns about AI security and the risks of advanced AI systems.