Europe's flagship artificial intelligence deadline arrived on 2 August. It requires machines to be labelled. It does not require anyone to show that the machines are right. Four developments in the two weeks around it show what that distinction costs in life sciences.

Three pharmaceutical companies have now claimed the largest AI machine in the industry. In the same fortnight, the first honest scoreboard for AI-discovered drugs showed the advantage thinning out at exactly the point where independent checking begins.

In one fortnight the AI tools got sharper and the deals got bigger. Not one of them made a single AI decision easier to verify. That gap is the whole game.

In two weeks, three drugmakers paid billions for the ability to generate proprietary biological data. Generating data faster is not the same as proving it is correct, and that gap is about to become a regulatory problem.