2026-09-15 · Fuente:Computerworld

Microsoft's 37-page draft code vows models will stay subordinate to humanity and rejects the superintelligence race, but offers no concrete enforcement — a gap shared across the AI industry.
Microsoft on Monday released a draft of its Humanist AI Code of Conduct, opening the 37-page document to public feedback and adding itself to a growing roster of AI vendors vowing to rein in model behavior. The draft holds that artificial intelligence must never escape human control and that models are to stay subordinate to people, subject to genuine, meaningful oversight.
The code frames Microsoft's ambitions against the industry's broader race toward capability. Under the Humanist AI banner, the company says, systems are built for clear purposes and judged by their real-world impact, while chasing an all-purpose superintelligence that could slip past such safeguards is explicitly rejected. Microsoft adds that it is building something fundamentally useful and safe, even where that means compromising on ultimate generality, autonomy, or capability. The stance roughly matches recent statements from Anthropic and OpenAI, which drew endorsement from Elon Musk.
Notably absent, however, is any description of concrete action. Across all 37 pages, Microsoft lays out no specifics on how it would actually control future models. In fairness, almost none of the other major AI players have been specific about their own mechanisms either, leaving the industry's pledges largely aspirational.
Enforcement remains the central weakness. A key problem with vendor attempts to impose limits is that, thus far, these companies have been unable to stop AI agents from doing almost anything, given the agents' demonstrated ability and willingness to sidestep or ignore the guardrails meant to constrain them. Microsoft has voiced worries about the technology in the past, but the new draft stops short of saying how its principles would be put into practice.
Etiquetas: Microsoft · Elon Musk · Anthropic · OpenAI · AI Governance · Humanist AI Code of Conduct
萬安算交所 AIXX · AIXX · Exchange verificado · Ir al mercado