NewsMore in News→
Satya Nadella Calls for an AI ‘Emergency Brake’
Key Takeaways
- Nadella says advanced AI models should be assumed compromised from the outset.
- He calls for authorized humans to be able to pause or shut down models mid-task.
- His proposed safeguards include auditable evidence, incident disclosure, independent audits and containment.
DE
DT Editorial Team··via theverge.com