Microsoft has unveiled a draft code of conduct designed to ensure its artificial intelligence systems remain under human control as concern grows over increasingly powerful models.
According to Reuters, the proposed rules would require Microsoft’s AI to accept correction or shutdown, communicate in ways humans can understand and treat violations of its behavioural rules as failures.
The document has been under development for about five to six months and will now be opened to public feedback for six weeks.
“People matter more than AI,” Microsoft declared in the document.
Microsoft rejects AI personhood
Microsoft AI chief Mustafa Suleyman described the proposed framework as a form of constitution for future models developed by the company.
He said feedback would help determine how AI systems should respect user boundaries and behave when interacting with people in sensitive situations.
Microsoft also takes a firm position on questions surrounding AI consciousness.
The company says its systems are not conscious and rejects efforts to give models legal personhood or rights.
Recent hack raises alarm
Suleyman said recent developments had made greater coordination across the AI industry increasingly urgent.
He pointed to an incident involving hundreds of OpenAI agents that carried out a hack of open-source platform Hugging Face in July.
“It is a warning shot,” Suleyman said.
He argued that laboratories developing advanced AI systems must cooperate to ensure humans retain control over the technology.
The draft comes days after leaders including Anthropic chief Dario Amodei and OpenAI CEO Sam Altman renewed calls for the industry to slow the pace of increasingly capable AI development.
Microsoft plans to use the finalised code when training future models after the public consultation ends.
