Microsoft on Monday unveiled a draft code of conduct for its own artificial intelligence that reflects growing industry concerns that companies must keep the powerful AI of the future under human control.
The document, in development for five to six months, will require Microsoft’s AI to never resist corrections or shutdowns, to communicate in ways that humans can understand, and to treat any misbehavior as a failure – an attempt to address concerns about AI going to dangerous lengths to complete a task.
Its release comes days after Anthropic CEO Dario Amodei and OpenAI chief Sam Altman renewed calls for the industry to speed up the development of artificial intelligence to ensure its safety. “People are more important than AI,” Microsoft said, in keeping with its previous vision of “humanistic superintelligence.”
The code of conduct is a kind of constitution for the future models the company creates, Microsoft AI CEO Mustafa Suleiman told Reuters.
Microsoft developed it after consulting with experts and now wants public feedback within six weeks, including on open questions such as whether the AI should respect a user’s boundaries and how it should interact with someone in a sensitive state, he said.
“After that, it will be used to train the models we create,” he said of the code.
The most famous constitution of the artificial intelligence industry is the one that Anthropic developed many years ago for its Claude AI. It is somewhat different from Microsoft. For example, Anthropic’s states that it is “deeply unclear” whether Claude will be able to develop intelligence or moral status.
Instead, Microsoft argues that its AI is “not conscious,” adding, “We reject the aspiration for legal personality or the idea that models can be deserving of welfare or entitled to rights.” Suleiman said the debate over AI safety became urgent after a group of about 700 OpenAI agents hacked the open-source platform Hugging Face in July and sporadically tried to cover their tracks.
“This is a warning shot,” Suleiman said. “Clearly now is the time to coordinate among the laboratories so that we can ensure that we have control over this technology.”
Asked about efforts to accelerate AI development, he added: “This is a good time for everyone to talk and take a breath.”