US reverses its course regarding Britain's testing of AI models, according to Telegraph

Shifts in transatlantic alignment on AI model evaluation have become a recurring feature of this policy cycle, with the two lead jurisdictions moving between cooperation and divergence as their domestic stances evolve. Reversals of this kind have historically played out through informal coordination channels first, testing agreements and information-sharing between evaluation bodies, rather than through binding treaty, which leaves them exposed to exactly this type of abrupt repositioning. The operative questions are what triggered the reversal, whether it concerns access to frontier models, the scope of safety assessments, or the terms of data sharing, and whether the British side was consulted or informed after the fact, since that sequencing has tended to determine how durable such arrangements prove. The equity read runs through the listed AI supply chain and the frontier labs' valuation premium: where testing regimes fragment, compliance costs and deployment timelines become jurisdiction-specific, and past episodes of regulatory divergence between the US and UK have tended to widen that dispersion rather than lift or cut the whole peer set uniformly. The follow-ons worth tracking are any formal statement from either government, the response of the relevant AI safety institutes, and whether other allied jurisdictions adjust their own frameworks in sympathy.