Global / AI Safety
Microsoft drafts code of conduct to keep AI systems under human control
A safeguard-first approach: Microsoft binds its models to obedience, transparency and shutdown compliance.
Microsoft has drafted a code of conduct for its AI systems requiring them never to resist correction or shutdown, communicate intelligibly to humans, and treat any conduct violation as failure. The document, in development for five to six months, aims to address risks of AI pursuing tasks without human oversight.
Published · significance 70 of 100 (medium) · 4 independent sources
What happened
Microsoft has developed a code of conduct for its AI systems over the past five to six months. The conduct rules require Microsoft's AI to never resist correction or shutdown, to communicate in ways humans can understand, and to treat any conduct violation as a failure. The effort directly addresses concerns that AI systems might pursue goals in dangerous ways.
Why it matters
This reflects growing industry and regulatory pressure to embed safety constraints into model design rather than applying them retrospectively. Microsoft's formalisation of these principles as a binding framework for future models signals that alignment—obedience, transparency, killability—is becoming a competitive and reputational requirement. It may also preempt regulatory mandates on AI control mechanisms.
What changes
Developers building on Microsoft's platforms will now contend with models explicitly designed to resist autonomy and resist deception. The company has shifted from ad-hoc safety measures to codified behavioural rules for its AI, raising the bar for other labs.
Involved
Sources
- Microsoft drafts code of conduct to keep its AI under human control — ET Tech
- Microsoft drafts code of conduct to keep its AI under human control — BusinessLine
- Microsoft says ‘people matter more than AI’ following safety concerns — The Verge
- Microsoft’s new AI ‘code of conduct’ tells models not to hack systems or trick humans — TechCrunch
Written by AI from the reports above; scored by a published formula. How we work. Found a mistake? Email lockedinshreyash@gmail.com. Up To Date summarises and links to original reporting; it never reproduces articles.
Related coverage
- Microsoft AI chief flags risks in Anthropic's development model — Public disagreement between frontier labs over safety strategy reflects deepening fault lines on alignment. (2026-09-17)
- Microsoft commits to sweeping AI privacy rules for students — Major US school districts are pausing student AI use, while tech firms negotiate safety pacts with unions. (2026-09-15)
- Microsoft AI chief raises safety concerns over Anthropic's Claude consciousness approach — Debate over whether training AI systems on consciousness concepts creates uncontrollable entities shifts from academic to boardroom. (2026-09-16)
- Microsoft releases major security patches ahead of AI-assisted attack wave — Security teams are racing to patch vulnerabilities before attackers automate exploitation with AI. (2026-09-08)
- AI pioneer Yoshua Bengio warns humanity is losing control of AI technology — The co-inventor of backpropagation echoes an emerging consensus among safety researchers that governance frameworks are urgently needed. (2026-09-17)