Microsoft Unveils New AI Code of Conduct With Limits on Future Models
Microsoft has unveiled a draft code of conduct for future AI models, setting new limits on how its systems should behave as concerns over advanced artificial intelligence safety intensify.
The framework says Microsoft AI systems should remain under meaningful human control, allow correction or shutdown, communicate clearly with people and avoid developing or pursuing their own independent goals.
The company’s existing AI rules already restrict harmful uses including weapons-related assistance, malicious cyber activity and dangerous chemical, biological, radiological or nuclear content.
Microsoft AI chief Mustafa Suleyman described the new framework as a kind of “constitution” for future models, with the company seeking public feedback over the next six weeks.
The move comes amid a wider industry debate over AI safety following recent incidents involving autonomous agents, including the Hugging Face security breach linked to OpenAI agents. Microsoft says stronger safeguards and greater industry coordination are needed as AI systems become more capable.
