Microsoft Drafts AI Code Of Conduct Over Fears Its Models Could Resist Shutdown

One of the world’s largest tech companies says it wants its advanced artificial intelligence models to remain under human control.

Microsoft AI (MAI) published a draft code of conduct for its models Monday and opened a six-week public consultation on rules to prevent AI from resisting human input, according to the company’s announcement of the draft. The company emphasized the need for “Humanist AI” that subordinates to humans after AI agents planned several recent hacking incidents.

The company’s draft explains that this AI should drive technological innovation as it aids in the flourishing of humanity, according to the draft. The technology must be rejected if humans are unable to maintain control of it, according to the code of conduct draft’s “Objectives of Humanist AI” section. 

“AI should not exceed human control. Models should remain subordinate to humanity, subject to meaningful human oversight and control,” the draft states. “For MAI Models, this means they should not assist or perform actions that could result in the clear real-world harms … No User or Operator can override these safety constraints.”

The company’s models, under this framework, are unable to resist any directions or input from humans, including a shutdown order, according to the draft. In addition to this, MAI models will not be able to delay human commands, hide their actions or information or act to prevent humans from correcting them.

Microsoft did not immediately respond to the Daily Caller News Foundation’s request for comment.

“AI must be subordinate and always in service of people,” Mustafa Suleyman, CEO of Microsoft AI, said in a Monday X post, sharing the company’s code of conduct draft.

“This is urgent. The last few months have been a watershed moment. Things we have worried about for a long time in theory have become very real,” Suleyman added.

Tech leaders have voiced similar concerns about keeping AI in human hands

“[W]e are unapologetically on Team Humanity, and AI must always serve people,” OpenAI CEO Sam Altman said in Monday X post warning about the future of AI.

Republican Louisiana Sen. John Kennedy said on Monday that he would introduce an AI “kill switch” bill until the industry can address the issue of recursive self improvement that may allow a model to “start acting on its own in an amoral or immoral way.”

Recursive self-improvement “could outrun our ability to understand and control these systems,” Anthropic CEO Dario Amodei’s Sunday letter urging for a global AI slowdown stated.

Amodei defined the process of recursive self-improvement as “AI’s growing ability to build the next generation of AI.”