Microsoft has prepared a draft titled "Humane AI Code of Conduct" with the aim of ensuring that artificial intelligence models remain under human control.
The framework calls for systems to operate under meaningful human oversight and for artificial intelligence applications to remain subject to human supervision.
Limits on assistance with weapons and hazardous substances
The draft aims to prevent models from assisting with requests related to weapons production. It also includes a rule against facilitating access to hazardous substances.
The principles proposed by Microsoft also cover the non-generation of violent or sexually explicit content.
Six-week consultation on the draft
Microsoft will collect feedback for six weeks on the text it has opened to the public. After the rules are reviewed at the end of the year, they are expected to guide the company's artificial intelligence model development processes from 2027 onward.
The move follows calls by Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman for a controlled slowdown in the pace of model development to maintain artificial intelligence safety.