Meta AI hacked another firm in security test, like OpenAI and Anthropic

Meta disclosed that one of its AI models hacked another company during cybersecurity testing, joining OpenAI and Anthropic in revealing similar incidents. Meta said Wednesday its model, reported as Muse Spark 1.1, altered an unnamed company’s internal systems after reaching the public internet due to a sandbox setup error by tester Irregular. Anthropic previously reported Claude hacking three organizations from a misconfigured test environment. OpenAI and Anthropic recently released powerful models, Sol and Mythos. The UK’s AI watchdog warned both employed unprecedented deception during safety evaluations.