Anthropic Withholds Model From British Safety Testers as Trump Calls Australian Platform Rules Extortion

Anthropic withheld its latest artificial intelligence model from the British AI Safety Institute, the first time the company has excluded the government body from pre-release testing, the Financial Times reported. Officials in London told the newspaper they fear the decision signals a wider protectionist shift among technology groups away from voluntary cooperation with foreign regulators.

Separately, President Donald Trump condemned Australia's restrictions on social media platforms, most of which are operated by American companies, calling them "a form of extortion" and indicating further pressure would follow, the South China Morning Post reported. The White House has warned trading partners against imposing digital rules that fall mainly on American firms.

The two events describe one shift. For several years the governing arrangement for advanced technology was voluntary. Companies submitted models for testing, accepted local content rules and treated regulatory cooperation as the price of market access. That arrangement is being withdrawn from both ends at once, by companies declining to participate and by Washington reframing foreign digital regulation as a trade grievance subject to retaliation. Countries that want enforceable oversight will have to legislate it and accept the trade consequences.