Archive · 12 Oct 2026
Nadella Urges Companies to Build Emergency AI Shutdown Safeguards
Satya Nadella says companies should be careful when using powerful AI systems.
He says they should plan for the possibility that an AI model could be compromised.
Companies should not depend only on what AI developers promise.
They should create ways for authorised people to pause or stop a model.
This should work even while the model is doing a task.
Nadella compared the safeguard to an emergency brake.
It would help people stop the AI if it behaves unexpectedly.
Microsoft Chairman and CEO Satya Nadella urged companies to prepare for advanced AI systems to be compromised.
He said organisations should treat powerful AI models as potential insider threats.
Nadella urged companies not to rely solely on assurances from AI developers.
He called for safeguards that let authorised personnel pause or shut down models while they are working.
He compared the safeguard to an emergency brake that can stop a model mid-task.
- Who
- Microsoft Chairman and CEO Satya Nadella.
- What
- He urged companies using advanced AI to plan for compromised models and create ways to pause or shut them down.
- When
- Not specified in the article.
- Where
- Not specified; Nadella shared his comments in a post on X.
- Why
- To allow authorised personnel to contain AI models if they are compromised or behave unexpectedly.
- Person
- Satya Nadella, Microsoft Chairman and CEO
- Topic
- Safeguards for advanced AI systems
- Recommended assumption
- A model could be compromised
- Recommended safeguard
- Authorised personnel should be able to pause or shut down a model during a task
- Analogy
- An emergency brake
- Source of comments
- A post on X