Anthropic says Chinese AI companies have been using Claude to help train competing models through what it calls “distillation attacks.”

The basic idea is simple: send large numbers of prompts to a powerful AI model, collect its answers, and use those outputs to help train another model.

Model distillation itself is a legitimate and widely used machine-learning technique. The controversy is over Anthropic’s allegations about how its systems were accessed and whether competitors violated its terms and safeguards to obtain the data.

Today’s Daily Dose comes from Plac Nowy in Cracow’s Kazimierz district, home of the Okrąglak and its famous zapiekanki.