UN warns on AI risk, regulation for Nvidia, Microsoft

Turk said he shared industry concerns that advanced AI could pose an existential risk and called for urgent safeguards before it is “too late.” He also pressed for international coordination on AI rules and renewed the push for a global governance framework, including the UN’s first high-level meeting on AI this year.

The comments reflect a widening split between the commercial drive to deploy more capable systems and the policy push to slow or constrain them. In filings, Microsoft, Nvidia, Meta and Alphabet have all flagged the risk that AI could trigger legal liability, regulatory action, data privacy problems, product failures and restrictions on frontier model development or deployment. That makes governance no longer a side issue but a direct input into margins, capital spending and valuations.

For the market, the issue is not whether AI remains a growth engine, but how much of that growth survives under tighter rules. Nvidia’s stock, which has been buoyant on demand for AI infrastructure, has also been volatile around policy and export-control risks, while Microsoft and Alphabet face the prospect that safety requirements could raise costs and slow commercialization. The sector’s recent rally assumes continued rollout of larger models, more data-center buildout and rising monetization; heavier oversight would not necessarily stop that cycle, but it could compress returns and lengthen payback periods.

Turk’s call for an immediate ban on autonomous weapons adds another layer to the debate. Military use cases are increasingly part of the AI conversation, and the prospect of restrictions on systems that can decide to kill without human involvement could spill over into broader rules on model controls, testing and accountability. That would be especially relevant for firms building foundational models and the semiconductor suppliers underpinning them.

The bullish case for the industry is that formal rules may legitimize AI’s use and reduce the risk of a public backlash after a major failure. The bearish case is that regulators, alarmed by the pace of progress, could impose costly guardrails just as companies are spending heavily to secure a lead in the next generation of models.

Either way, Turk’s warning underscores that AI policy is now a market variable. The next catalysts are likely to be new national rules, international coordination efforts and further company disclosures on safety, liability and compliance costs.