Liability
for incorrect
AI results.
Liability
for incorrect
AI results.
from
What happens if an AI portal publishes false information about your company? Why even a disclaimer in the terms of use offers no protection.
What had happened?
A medium-sized company that builds and sells conservatories and patio roofs found a frightening report about itself on a business information portal. The portal had automatically retrieved data from the commercial register and incorrectly assigned an entry.
The message stated that the company was to be deleted due to lack of assets. That was simply wrong. The notification belonged to another company with a similar name. An AI system had mixed up the data during automated processing.
The portal appeared in second place in searches for the company and was therefore particularly visible. Anyone who googled the company immediately saw that it was supposedly on the verge of insolvency.
The company demanded the immediate deletion of the report and a declaration to cease and desist. The portal deleted the report immediately, but refused to sign the declaration. This led to a lawsuit.
What did the court decide?
With Judgment of 29.02.2024 – Ref. 6 O 151/23 the Kiel Regional Court ruled in favor of the company and condemned the portal operator.
The operator is liable – even if not at fault
The court classified the portal operator as a direct interferer. This means that anyone who uses an AI to process and publish company data is responsible for the result – regardless of whether they themselves were actively involved in the incorrect report.
The defendant cannot claim that it was not involved in this automatic process because it deliberately used artificial intelligence to answer its users’ search queries, which was inadequately programmed in cases such as this.
The key idea is that anyone who voluntarily activates an AI must accept responsibility for the results of this AI. This also applies if the AI was inadequately programmed and produces errors.
Anyone who bundles content makes it their own.
The portal had pointed out in its terms of use that all information is obtained fully automatically and potentially incorrectly. The terms and conditions also contained a general disclaimer.
The court did not accept either. The decisive factor was that the portal brings together mandatory publications from various registers, links them and presents them in a structured manner. In doing so, it assumes recognizable responsibility for the content of the information presented. The court referred to this as “attribution”.
Inaccurate AI results?
Here’s how to
best respond to
false
statements made
by AI!
Overviews and summaries generated by artificial intelligence often appear prominently at the top of search results and are perceived by users as reliable summaries.
If the statements contained therein are false or defamatory, the economic and reputational consequences can be significant. Several courts have ruled that providers are liable for false statements in such AI-generated representations.
Timely and professional advice, along with swift action, are crucial for addressing such legal violations.
- Preserve evidence immediately: AI responses can change at any time. What appears in an AI overview today may look different tomorrow or have disappeared entirely. Document the exact search query, date, and time; the complete AI response; a screenshot of the entire page; and the sources displayed along with their links. Without careful documentation, it will be nearly impossible to prove later what was actually displayed.
- Have a legal review conducted: Not every unfavorable or incomplete representation automatically gives rise to legally enforceable claims. A successful legal action always requires an actual violation of the law. Inaccurate or incomplete wording is not sufficient in individual cases. Have it reviewed to determine whether the AI statement is demonstrably false or misleading and infringes upon a protected legal right.
- Understanding Your Rights: False factual claims made by AI can give rise to claims for injunctive relief, removal, correction, and, where applicable, damages. Depending on the specific case, various legal grounds for claims may apply, including claims for infringement of a company’s right to privacy, damage to creditworthiness, and claims under competition law pursuant to the UWG. In the case of natural persons, such as managing directors or freelancers, additional claims under personality rights and data protection law may also apply.
- Develop a Strategy: Sometimes an incorrect AI response is based on a flawed or outdated source. In such cases, it may make sense to address the source information first, as correcting the source data can influence the AI’s response. In addition, depending on the circumstances, options include a cease and desist letter, a preliminary injunction, or a lawsuit. The appropriate strategy depends on the nature of the misrepresentation, the platform in question, and the extent of the damage, and should be developed with the assistance of a lawyer.
Conclusion
Those affected by violations of personal rights or other violations based on incorrect AI results can assert claims against the portal operator who adopts these as their own. Standard disclaimers in terms of use are not sufficient.
For portal operators, this results in enormous risks and high duties of care when using artificial intelligence.
We are happy to
advise you about
AI!
