When AI chatbots lie, companies are liable: Landmark German ruling on the attribution of AI statements
Dr. Martin Eckert
Legal Partner
Higher Regional Court of Hamm, Judgment of May 12, 2026 – I-4 UKl 3/25 (not yet final)
The German court ordered a cosmetic surgery clinic to cease and desist because its chatbot made false statements about the doctors. The clinic used an AI chatbot on its website to schedule appointments and answer questions. When asked, the bot replied that the two managing directors were “specialists in plastic and aesthetic surgery” and “specialists in aesthetic medicine”— titles that the two doctors did not actually hold. Although the chatbot was shut down, the company did not issue a cease-and-desist declaration. The Higher Regional Court of Hamm therefore ordered the company to cease and desist.
The court deemed the chatbot’s incorrect responses to be misleading commercial acts by the company itself under Section 5(1) and (2)(3) of the German Unfair Competition Act (UWG).
The Senate clarified two points in this regard:
The chatbot is not to be regarded as a “third party” within the meaning of competition law. It is part of the company’s business organization — its statements are attributed to the operator.
Even correct programming does not preclude liability. Even if the chatbot was trained exclusively with accurate data sets, the company is responsible for the false statements.
This means that companies cannot rely solely on compliance with unfair competition law’s duty of care when using AI chatbots.
The ruling thus takes a clear stance on the long-standing question of who is responsible for hallucinated or erroneous AI responses — the company, the AI provider, or “no one”? According to the Higher Regional Court, at least under unfair competition law: whoever uses a chatbot on their site bears the risk of hallucination themselves.
Although the ruling was issued in the context of competition law, the underlying principle — “Whoever uses the bot is responsible for its output”— may be applicable to other areas of law:
Data protection law (GDPR): Even in the case of automatically generated personal statements (e.g., false information about customers, applicants, employees), the controller may be responsible for their accuracy. In addition, information obligations under Art. 13/14 GDPR and — in the case of automated individual decisions — Art. 22 GDPR apply.
AI Regulation (AI Reg/AI Act): Article 50(1) of the AI Regulation requires, effective August 2, 2026, that users be transparently informed that they are communicating with an AI.
E-commerce / Consumer Law: Incorrect statements regarding products, prices, or delivery can, in addition to violations of information obligations that may result in cease-and-desist orders, also lead to contractual obligations, warranty claims, damages, or fines (UWG, PAngV).
General Liability Law: Third-party personality rights, copyright, and trademark law may also be affected by chatbot outputs and give rise to claims by the affected parties.
In practice, therefore, when using chatbots and AI assistants, it is advisable to implement both technical and organizational measures to closely monitor the admissibility and accuracy of the output. This is most likely to be achieved through a combination of several measures (e.g., topic filters, guardrails, escalation to human review, ongoing monitoring, and spot checks). Due to the particular significance of the issues regarding the attribution of false statements made by chatbots, the Senate has granted leave to appeal to the German Federal Court of Justice.
(Author: Dr. Marian Klingebiel, UNVERZAGT Law)
The Swiss Perspective on Liability for AI
Swiss courts are likely to reach a similar conclusion based on the Swiss Federal Acton against Unfair Competition (UWG).
In general: The use of chatbots and AI assistants entails risks that (Swiss) companies must adequately manage and for which they may be held liable, depending on the circumstances. The same applies to AI: cura in eligendo, instruendo, and custodiendo. When using AI, the company must carefully select the AI tool, organize its use appropriately, and monitor the output.
