← Back to live feed1 story
Microsoft CEO Satya Nadella said human-controlled 'emergency brakes' are required to shut down agentic AI models mid-task, and urged organizations to treat frontier models as insider threats from the start. He called for security permissions enforced outside the model, supported by independent audits and tamper-proof logs, and said users cannot outsource responsibility for an AI's actions to the model provider.
Sign in to suggest edits
Key sources
- SOURCE@satyanadella“https://t.co/MjwFy73LOF”x.com
- SUPPORT@business“create an “emergency brake” system to prevent agentic models from going rogue”x.com
- SUPPORT@choblin29“The most trustworthy Super Intelligence system will not be the one with the model we trust most. It will be the one that enables us to trust the model the least.”x.com
- SOURCEhuggingnewshuggingnews.com