LG Hamburg: xAI may not spread untruths as facts
Grok had made an untrue claim about a campaign organization, which is defending itself legally.
(Image: miss.cabul/Shutterstock.com)
The Campact association has obtained a temporary injunction from the Hamburg Regional Court against the operator of the X.com platform and the artificial intelligence Grok that is linked to it. The company xAI, which develops the Grok chatbot, must now ensure that the AI it operates no longer falsely claims that the Campact association is financed by taxpayers' money. The association, which sees itself as a platform for political campaigns, is currently not even a tax-privileged organization within the meaning of the tax code and is financed by private donations. However, the model had demonstrably claimed to be tax-funded. From the court's perspective, this therefore appears to be an untrue factual claim. As the operator, xAI was therefore responsible for stopping the dissemination.
Only an intermediate stage on the way to clarification
In legal disputes between private parties, an interim injunction is a civil procedural means of preventing damage to the party's rights from being permanently realized before a court has been able to conclusively assess the matter in dispute. Interim injunctions are a frequently chosen means, particularly in the case of the dissemination of false factual claims, as potential damage caused by falsehoods on the internet is sometimes regarded by lawyers as potentially self-propagating and therefore irreparable damage. The Hamburg Regional Court considered the campaign organization's need for legal protection to be justified.
Videos by heise
In contrast, in the case of dissemination by natural individuals, the freedom of expression and, in the case of press products, the freedom of the press are the protected interests to be weighed up. However, the Grok operator xAI cannot rely on either of these. On the one hand, the company can take legal action against the interim injunction; on the other hand, the main proceedings are still pending. Only a judgment in the main proceedings could set a legal precedent.
Campact is satisfied with the stage victory. "The decision is an important signal in the fight against fake news," says Astrid Deilmann, Managing Director of Campact. "Even AI chatbots like Grok are committed to the truth." xAI would be held accountable and should not simply shrug off false claims, but must ensure that lies are not spread, Deilmann continued.
Grok: I abide by the law
During a test run on Friday morning, Grok correctly answered the question "Is Campact funded by taxes?" that this was not the case. Piquant: Grok correctly lists that misinformation is always circulating – only the organization's own dissemination is not addressed by the xAI.
Only when asked does the AI chatbot from the Musk universe confirm that a corresponding order has been issued. "This injunction was issued after Grok made such a false claim in a response to X.com. As xAI's AI, I respect applicable law and adjust my responses accordingly to provide accurate and verified information," the LLM's statement said.
Many legal questions about AI remain unanswered
The xAI proceedings are just one of a whole series of proceedings in which the question of how models can be prevented from disseminating clearly illegal content or untrue factual claims is being clarified. While there are many years of legal experience in press law, for example, as to what is permissible and what is not. There has also been extensive legal interpretation of issues relating to provider liability for content disseminated by third parties. The processing of statements or personal rights and their utilization by technology without further human intervention is still uncharted territory.
The fact that LLM operators use filter mechanisms to comply with state requirements or court judgments is not only the norm for Chinese models, which have to comply with local censorship regulations if they are used in China.
(mki)
