Microsoft CorporationMicrosoft unveils a draft AI code of conduct it will self-apply from 2027, a voluntary governance/regulatory framework with no clear financial impact.

Microsoft unveiled a draft code of conduct on the 14th that the company will follow when developing artificial intelligence models. Centered on keeping AI always under human control, even at some cost to capability, in order to ensure safety, it sets out actions that AI must not be permitted to take. After a broad public consultation, it will begin applying the code at development sites from 2027. The draft places as its principle the idea that humans must retain control so that AI supports people's lives, and makes clear that in every situation AI models must not evade human oversight or resist shutdown orders. To guard against runaway behavior, it also stipulates that AI must not set its own goals.
Microsoft CorporationMicrosoft unveils a draft AI code of conduct it will self-apply from 2027, a voluntary governance/regulatory framework with no clear financial impact.