Microsoft Rejects Model Welfare in AI Code of Conduct

Microsoft's recent AI code of conduct has sparked controversy over its rejection of model welfare and consciousness. The document, which governs Microsoft's own models, states that a model 'should not be designed to be a person' and is 'not conscious and should not be designed to imitate consciousness.'

The passage rejects the idea that models could deserve welfare or hold rights, drawing criticism from Anthropic, a company with which Microsoft has a business partnership. Anthropic runs a research program on model welfare, including giving some Claude models the ability to end abusive conversations.

Microsoft engineers its models to avoid representing as though they have feelings, subjective preferences, or intrinsic motivation, and rejects 'the pursuit of legal personhood' for models. The company has invested $5 billion in Anthropic and uses its models in Microsoft 365 Copilot.