A few months ago, the UK government was preparing a concise piece of legislation that would require AI companies to submit large models for evaluation by the UK’s AI Security Institute before public deployment. It was a targeted, modest proposal — not the sweeping framework that Brussels built, but something deliberately narrow: a safety gate for frontier AI. The kind of thing a government does when it wants to be seen as taking risk seriously without terrifying investors.

Now there is no bill. Science Minister Patrick Vallance recently told MPs, plainly, that there is “no bill at the moment.” What had been described as a pre-Christmas release has been pushed to summer at the earliest, and sources within government are characterizing it as “properly in the background.”

The reason, as with most things in Westminster these days, is Washington.

Since Donald Trump’s return to office, the US has dismantled the AI safety architecture Biden built. The AI Safety Institute — the US counterpart to the UK’s AISI — has seen its leadership resign. The outgoing administration had positioned it as a cornerstone of global AI governance. The incoming administration called it bureaucratic overreach and quietly moved to starve it of relevance. VP JD Vance used a Paris AI summit to publicly criticize Europe’s regulatory ambitions, and the UK — which had been quietly signaling willingness to sign the resulting Paris declaration — notably declined. Sixty-six nations signed. The UK did not.

What this tells us is that the UK’s AI regulatory ambitions are now subordinate to its transatlantic investment strategy. Peter Mandelson, the UK’s ambassador to Washington, has reportedly been focused on positioning Britain as the preferred European destination for US AI capital. That’s not a regulatory posture — it’s an economic one. And when those two objectives conflict, the economic one wins in this government.

The timing is instructive. The government is also under pressure over copyright. Artists including Paul McCartney and Elton John have raised serious objections to proposals that would allow AI companies to use online materials without seeking permissions. These are politically expensive fights — the entertainment industry is vocal, organized, and disproportionately represented among the voters who fund the Labour Party. But they’re also secondary to the industrial strategy. The government’s silence on the copyright question mirrors its silence on the AI bill: nothing concrete, nothing denied, just a slow disappearance into the background.

This is what regulatory abdication looks like when it’s not called that. The UK isn’t repealing rules it hasn’t passed. It’s simply not passing them, and hoping the market interprets inaction as friendliness. Whether that calculation pays off depends entirely on whether US AI companies see the UK as a beachhead into Europe or just another market. Right now, the signals are pointing toward the former. That won’t last if the EU’s AI Act starts producing enforceable judgments and companies need a compliant jurisdiction to operate in. Brussels regulates. London waits. We’ll see which model produces more AI investment by the end of the year.