Microsoft's new rules bar its models from hacking systems or deceiving humans, framing AI as a support tool rather than a replacement.
Microsoft released a 37-page 'code of conduct' for its AI models, laying out principles that include supporting rather than replacing humans and specific safety constraints against hacking or tricking users. The move lands amid a broader industry debate over AI safety sparked by Anthropic's call to slow development.
The document positions Microsoft as taking a public stance on AI governance ahead of any binding regulation, effectively setting a company-level standard that competitors will be measured against.
Self-regulation from the largest enterprise AI vendor sets a de facto industry baseline before lawmakers act. Enterprise buyers evaluating AI vendors now have a public benchmark to hold Microsoft, and by extension its competitors, accountable to.
The daily signal, curated. Get it in your inbox.
Subscribe on LinkedIn →