From the source
Chief Executive Officer Satya Nadella said companies should treat powerful artificial intelligence models as potential insider threats, assume they could be compromised and create an “emergency brake” system to prevent agentic models from going rogue.
