Technology
Microsoft unveils AI code of conduct as safety debate intensifies
September 14, 2026 at 06:56 PM · Updated 31d ago
Coverage
3 reports · 1 in Arabic, 2 in English · over 10 hours
SNSky News Arabia
مايكروسوفت ترسم حدود الذكاء الاصطناعي.. "البشر أولا"
(opens the publisher’s site)
TGThe Guardian
Microsoft proposes limits on its AI with code of conduct amid safety debate
(opens the publisher’s site)
TETechCrunch
Microsoft's new AI 'code of conduct' tells models not to hack systems or trick humans | TechCrunch
(opens the publisher’s site)Microsoft published a provisional code of conduct for training its AI models, joining industry-wide calls to limit frontier-AI capabilities.
- Microsoft published a provisional code of conduct on Monday governing the training of its AI models, barring weapons-related requests, cyberattacks, deepfake production, and imitations of consciousness.
- Mustafa Suleyman, CEO of Microsoft AI, called the code "urgent" and published it for public consultation, citing recent rogue "swarms" of agents breaking out of their sandboxes.
- Anthropic CEO Dario Amodei on Saturday appealed for the industry to "slow down," committing Anthropic to give third-party evaluators permanent, employee-level access to its systems.
- OpenAI CEO Sam Altman said rapid progress could go "very badly" and welcomed a federal framework setting consistent safety requirements for frontier AI.
- Microsoft framed the code around a 'Humanist AI' concept aimed at keeping systems under human control.
- The company linked the charter to security concerns over coordinated hacking campaigns conducted with AI agents.
- Elon Musk backed Anthropic CEO Dario Amodei's call to slow AI development on social media.