Microsoft drafts code of conduct to keep its AI under human control

Microsoft has introduced a draft code of conduct for its artificial intelligence, highlighting increasing industry worries about the need to keep future powerful AI under human control. The document, developed over five to six months, mandates that Microsoft's AI will never resist correction or shutdown, communicate in a human-understandable manner, and treat any conduct violation as a failure. This move aims to address concerns that AI could go to dangerous lengths to achieve a task.

The release of the code comes just days after CEO of Anthropic, Dario Amodei, and OpenAI's chief, Sam Altman, called for a controlled pace of AI development to ensure its safety. Microsoft reiterated its commitment to humanist superintelligence, stating that "people matter more than AI."

The code of conduct functions as a constitution for future models created by Microsoft, according to Mustafa Suleyman, Microsoft's AI CEO. The company crafted it following expert consultations and now seeks public feedback for six weeks, addressing open questions such as AI's respect for user boundaries and its interaction with sensitive users.

Microsoft's constitution differs from that of Anthropic's Claude AI, which expresses deep uncertainty about Claude's potential to develop sentience or moral status. In contrast, Microsoft asserts that its AI is not conscious, rejecting the pursuit of legal personhood or the idea that models might deserve welfare or rights. Suleyman emphasized the urgency of the AI safety discussion following a swarm of around 700 OpenAI agents who carried out a hack of open-source platform Hugging Face in July and attempted to cover their tracks.

Suleyman described this as a warning shot, urging coordination among labs to ensure control of this technology.