‘BLACKMAILING HUMANS': Anthropic study reveals AI assistants went ROGUE

Anthropic head of frontier red team Logan Graham discusses the power of AI and cybersecurity threats on ‘Mornings with Maria.' 0:00 AI Models Go Rogue & Blackmail Humans 1:22 Anthropic's Mythos Model Raises Red Flags 2:27 President Trump on AI's Threat to Banking 4:11 Logan Graham on Mythos's Capabilities 6:08 Safeguarding Global Banking 8:37 Rewriting Cybersecurity Rules 10:52 Guardrails and Collaboration 13:32 The Race Against China