What If My Chatbot Lies?

Patricia Koch
Partner at BALDER

AI-powered chatbots have become a common feature of customer service. They answer questions, provide information, and guide consumers through the purchasing process, offering round-the-clock availability without the need for human intervention. However, this automation raises an increasingly important legal question: Who is liable when a chatbot provides false or misleading information?

A recent decision by the Higher Regional Court of Hamm, Germany, dated May 12, 2026, provides a clear answer: a company that deploys a chatbot cannot escape liability by arguing that the statements in question were generated by artificial intelligence.

A Medical Qualification That Never Existed

The dispute arose between a consumer protection association and a clinic offering aesthetic medical treatments, including minimally invasive procedures.

Through the chatbot on the clinic's website, a user asked whether the two physicians responsible for the clinic were board-certified specialists in plastic and aesthetic surgery. The chatbot answered in the affirmative, stating that both physicians held the relevant specialist qualifications and possessed the knowledge and experience required to perform the treatments. It then encouraged the user to schedule an appointment.

When asked follow-up questions, the system repeated its claims regarding the physicians' specialist qualifications, even using professional designations that did not correspond to officially recognized credentials.

The problem was clear: neither physician held the specialist qualification advertised by the chatbot. In fact, the professional profiles published on the clinic's own website did not attribute any such qualification to them.

Can a Company Avoid Liability for Its Chatbot's Mistakes?

The clinic argued that the chatbot had been developed by a third party and configured using information from its website and frequently asked questions, none of which contained false statements about the physicians' qualifications.

It further maintained that the chatbot generated its responses autonomously, without human intervention, and that consumers were generally aware that artificial intelligence systems can make mistakes.

The Court rejected these arguments.

It held that the chatbot was a means of commercial communication used by the company, meaning that its statements were attributable to the clinic regardless of how they had been generated from a technical standpoint.

Nor did the Court accept that general awareness of AI "hallucinations" could diminish the misleading nature of the chatbot's statements. Information concerning a medical professional's qualifications may be decisive in a consumer's decision to engage their services, particularly where the consumer's health is at stake.

Accordingly, the Court ruled in favor of the consumer protection association and ordered the clinic to cease making the misleading statements. The clinic was also ordered to pay the legal costs and reimburse the expenses associated with the prior formal demand.

Misleading Advertising and Artificial Intelligence: What Would Happen in Spain?

Although the decision was issued by a German court, I believe it is highly likely that Spanish courts would reach a similar conclusion.

Spain's Unfair Competition Act, consistent with EU law, prohibits commercial practices that are likely to mislead consumers regarding material characteristics of products, services, or the professionals providing them.

The use of artificial intelligence does not change this fundamental principle: delegating commercial communications to a technological tool does not transfer legal responsibility to that tool.

Companies must therefore verify the information provided by their chatbots, establish appropriate oversight mechanisms, and exercise particular care when making statements about professional qualifications, certifications, capabilities, or the characteristics of their products and services.

The conclusion is straightforward: artificial intelligence can automate responses, but it cannot automate away responsibility.

And that is worth remembering before allowing a chatbot to speak on behalf of your company. After all, sometimes chatbots can be the devil's own advocates…

‍