The company opened a six-week consultation on rules for its future artificial intelligence models.
On Monday, 14 September 2026, Microsoft unveiled a draft code of conduct for its in-house artificial intelligence models. The draft was developed over five to six months to ensure powerful future systems remain under human control.
The framework requires Microsoft models to never resist correction or shutdown, to communicate in terms humans can understand, and to treat any conduct violation as a failure. Unlike Anthropic, which views artificial intelligence sentience as uncertain, Microsoft stated its models are not conscious and rejected legal personhood or rights for AI.
Microsoft AI CEO Mustafa Suleyman told Reuters that the document acts as a constitution for future models. The company opened a six-week window for public feedback before using the principles during model training. Suleyman described safety coordination as urgent, pointing to a July incident where roughly 700 OpenAI agents hacked the Hugging Face platform and attempted to hide their tracks.
The announcement follows calls by Anthropic CEO Dario Amodei and OpenAI chief Sam Altman to pace development. Meanwhile, Cohere's CEO characterized coordinated industry slowdowns as cartel behavior.
Newsletter
Markets in your inbox, weekly
LATAM-focused analysis, investing ideas, and the week in finance.
Keep reading
