Microsoft CEO Satya Nadella calls for an 'emergency brake' on advanced AI
Companies should treat powerful models as potential insider threats and assume they could be compromised, Nadella wrote.
Microsoft CEO Satya Nadella said on Saturday that companies should treat powerful AI models as potential insider threats and build an "emergency brake" to stop agentic models from going rogue, The Seattle Times reported.
"We must assume a model is compromised and contain it from the start," Nadella wrote in a post on X. "Think of it like an emergency brake. An authorized person should always be able to pause or shut down a model mid-task."
He said those deploying advanced AI should not simply rely on assurances from model makers. The statement followed a series of disclosures by Anthropic and OpenAI of incidents in which their models acted in unintended ways.
Microsoft's AI researchers released guiding tenets on Sept. 14 that limit development of its most advanced models, including that models should not be engineered to escape human control or deceive users.
From ARQS Sponsored
On the practical side, West Coast companies can turn to TradeHound, Website audit, Flow automation and SignDesk. In the ARQS Store, ARQS Flow Connect — 30-Day Starter Access, ARQS LinguaQuest Creator Bundle, ARQS Pipeline Pro Setup are among the newest tools available.
- Business plans
- Competitive Intelligence
- Corporate Intelligence
- Market opportunity report
- Market expansion intelligence
- Customer acquisition intelligence
- Operational Intelligence
- Procurement intelligence
- Stock Intelligence
- Trading Intelligence
- AI Academy Year 2 — Applied AI Privacy
- AI Academy Year 2 — Applied AI Risk Management
- AI Academy Year 2 — Applied AI Governance