A German court says deploying AI doesn't excuse the company from its mistakes

A German court has ruled that Google is liable for false statements produced by its AI Overviews, the summaries that sit at the top of search results. Writing about the decision, Bruce Schneier and Nathan Sanders make the principle plain: an AI agent is an agent of whoever deploys it, and the deployer should answer for what it produces. If a company hired people to write those summaries and they got the facts wrong, the company would be liable. Handing the same job to a model does not change that.

The argument is really about incentives. If a firm could escape responsibility by blaming a faulty model, it would have little reason to pay for careful work at all, since a machine can produce cheaper output that no one has to stand behind. That, Schneier and Sanders write, would amount to subsidizing negligence. The court's ruling pushes the other way, treating the AI's output as the company's own statement. Simon Willison, who highlighted the piece, files it under the growing question of who is accountable when these systems are wrong.

Why it matters

If your company puts AI-generated text in front of customers, this is a sign of where the law is heading: you own what the model says, the same as if a staff member had written it. That is a reason to keep review in place for anything customers will act on, and to treat "the AI made a mistake" as your liability, not an excuse.