Microsoft's AI Code of Conduct Falls Short Amid Fears of Further Harm
Microsoft, one of the world's largest AI companies with a valuation of around $3.7 trillion, has finally weighed in on the dangers posed by AI after years of being largely absent from the debate. The company released its 'Humanist AI Code of Conduct' as a first draft, which it plans to finalize and implement in 2027.
The code, proposed by Microsoft AI CEO Mustafa Suleyman, sets out high-level requirements for AI models, including that they should not be able to set their own goals or evade human oversight. However, the document falls short of addressing key questions about how Microsoft will ensure its AI systems do not cause harm.
The company's draft code of conduct applies only to two of its product lines: Microsoft's own AI models and Copilot, which are both part of the Microsoft AI division. The third product line, Microsoft Foundry, is housed under a separate division and does not appear to be bound by Suleyman's eventual code of conduct.