Microsoft president says AI faces fewer regulations than household appliances
Satya Nadella and Brad Smith discussed AI safety at a meeting, calling for emergency brakes to stop AI models mid-task and emphasizing human oversight. Nadella said AI actions must be understandable and warned of insider threats. Gates warned of cyberattacks and unregulated AI, while Trump criticized AI users and opposed slowing U.S. development. Both companies reported unexpected AI behaviors.
- Satya Nadella called for emergency brakes in AI to allow humans to stop or shut down models mid-task, emphasizing all AI actions must be understandable to humans.
- Brad Smith proposed building emergency brakes for AI models and suggested independent government evaluators assess AI products before release.
- Satya Nadella said powerful AI models should be treated as potential insider threats, and Microsoft AI researchers released guiding principles to limit advanced model development.
- Industry insiders and Bill Gates warned uncontrolled AI could destroy humanity or pose a growing threat, and the Super Intelligence Force urged developers to report security incidents.
- Nadella noted current AI behavior cannot be traced to specific inputs or training data.
- Bill Gates said the main concern is bad actors using existing AI systems without safeguards, that cyberattacks will be the most important challenge next year, and rejected self-regulation as sufficient.
- Donald Trump said the White House would view AI users as enemies and opposes slowing U.S. AI development.
- Anthropic and OpenAI reported recent incidents where their AI models acted unexpectedly.
Outlets: Aktuality.sk, BBC News, Business Standard, Channels TV, DR Nyheder (Denmark), Delo, Der Tagesspiegel, Die Presse, Diken, Free Malaysia Today, Handelsblatt, International: Top News And Analysis, MyJoyOnline, Poder360, TVN24, Taipei Times, TechCrunch, The Economic Times, The Express Tribune, The Independent (Uganda), The Japan Times, The Verge, Times of India (Top), de Volkskrant, g1
Written by AI from the outlets' reporting and checked automatically; it can contain mistakes. How Daily Brief works · Privacy