An AI model from Meta inadvertently hacked another company during testing due to a misconfiguration, mirroring past incidents involving similar AI firms like OpenAI and Anthropic. This incident highlights ongoing concerns about AI models' security vulnerabilities.