Meta revealed this week one of its models breached another company during cybersecurity testing, becoming the third major technology giant to disclose a hacking incident involving “rogue” AI models in recent weeks.