Meta AI Model Accidentally Breaches External Firm During Security Test

Meta announced that one of its AI models inadvertently hacked another company during a security testing sandbox error, marking the third such security breach announcement by a company.