Ang sinabi ni Nadella
Noong Oktubre 10, 2026, inilathala ni Microsoft CEO Satya Nadella ang sanaysay na "Models as Insider Risks in the Super Intelligence Era": dapat ituring ang advanced AI models, closed man o open-weight, bilang insider risk.
"Dapat nating ipalagay na compromised ang model at i-contain ito mula sa simula. Isipin ito na parang emergency brake."
Hindi ito panawagang itigil ang AI development, kundi control architecture.
Mga prinsipyo
- Paghiwalayin ang intelligence at authority (permissions sa labas ng model).
- Model diversity.
- Tamper-proof at nababasang record.
- Tuloy-tuloy na testing, independent controls at audit.
- Containment at mabilis na pagbubunyag ng insidente.
Para sa mga negosyo
- Minimal na credentials bawat agent.
- Policy sa labas ng model gamit ang gateway.
- Human approval sa kritikal na aksyon.
- Immutable logs.
- Nasubok na kill switch.
Mga pinagmulan
- CNBC (2026-10-10): https://www.cnbc.com/2026/10/10/microsoft-satya-nadella-ai-emergency-brake-safety.html
- The Verge (2026-10-10): https://www.theverge.com/ai-artificial-intelligence/1009337/satya-nadella-says-we-should-assume-all-ai-models-are-compromised
- Bloomberg (2026-10-10): https://www.bloomberg.com/news/articles/2026-10-10/microsoft-ceo-nadella-calls-for-emergency-brake-on-advanced-ai
- The Straits Times (2026-10-11): https://www.straitstimes.com/world/microsoft-ceo-nadella-calls-for-emergency-brake-on-advanced-ai
Gusto mo bang ilapat ang insider-risk controls sa iyong AI agents? Nag-aalok ang KGA IT ng libreng konsultasyon: kga-it.com.