German court holds Google liable for false AI Overview answers; here’s where Indian law stands
The decision came after two publishing companies in Munich filed a case, alleging that Google's AI Overviews were showing false claims about them.
5 Min Read
A German court has ruled that Google can be held directly responsible for incorrect answers in its AI-generated ‘Overview’ feature.
The decision came after two publishing companies in Munich filed a case, alleging that Google's AI Overviews, the short AI-written summaries you see at the top of search results, were showing false claims about them. The Regional Court of Munich has now issued a temporary injunction in the matter.
Publishers challenge AI overviews
The publishers claimed that Google's AI summaries were linking them to scams, subscription traps and other shady business practices. The companies said the claims were false and damaging to their reputation.
The court said that Google's AI Overviews cannot be treated like normal search results and may make the company directly responsible for false information.
AI overviews are not traditional search results
The court further said traditional Google Search only acts as an intermediary, showing links to third-party websites. But AI Overviews are different because they combine information and present it in Google’s own words. This gives users full answers without needing to visit source websites.
"The AI-generated summaries create independent, new, and substantive statements and therefore amount to Google’s own content rather than just third-party information," the court ruled. Judges said Google has full control over how the AI works, including the algorithms behind it. The court said Google is effectively the author and publisher of the AI-generated text.
What court found
The court found that the AI Overview made claims that are not even made in the search results. It also discovered that Google's AI Overview had mixed up information from different sources and, in some cases, created new and incorrect connections that didn't appear in any of the original websites.
Why Google is answerable
Google may not be protected under normal rules that shield online platforms as 'host providers' under the EU Digital Services Act, which usually limits liability for third-party content and requires only notice-and-removal action.
During the court hearing, Google argued that users should not fully trust AI summaries and should always check the linked source websites to verify the information. The company said people already understand that AI-generated content can be wrong, and therefore they should not rely on it blindly.
The court said AI Overviews function like standalone statements, similar to newspaper headlines or teaser lines that can still be legally challenged if they are misleading. "If AI Overviews are treated as publisher speech rather than intermediary indexing, every factual error becomes a potential liability event for Google," the court said.
The judges also said that even if users could check the sources later, that does not remove Google’s responsibility for what the AI already said. The court also warned that if AI answers are treated as "not fully reliable," then the feature itself would lose its value.
One analysis cited in the case suggested Google’s AI Overviews are about 91% accurate, but more than half of even correct answers do not clearly show supporting sources.
Earlier findings
The court also looked at earlier rulings by Germany’s Federal Court of Justice, which had given traditional search engines like Google some protection from liability.
Those earlier decisions said search engines are only ‘indirectly responsible’ because they help users find content from other websites. Courts also said search engines should not be forced to manually check every result, as that would make search services difficult to operate.
But the Munich court said this rule does not apply to AI Overviews.
It explained that a normal search engine only shows links to outside websites. However, AI Overviews read information from different sources, combine it, and generate new sentences in Google’s own voice which the court said they are creating new statements.
"Google can check those statements at least by comparing the underlying third-party websites with its own statements based on them," the court ruled.
The ruling also warned that this issue is not limited to Google. Other AI answer tools, including systems from OpenAI, Microsoft’s Bing, and Perplexity, could face similar liability questions if they generate answers in their own voice.
How social media liability works in India
In India, social media platforms are generally not responsible for user-generated content because they are treated as intermediaries, not publishers. If someone posts illegal content, such as copyrighted material or harmful content, the platform must act once it becomes aware of it.
The Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021, set strict rules for social media companies. These rules state that platforms must act quickly when illegal content is reported, follow proper monitoring and complaint systems and remove unlawful content when required.
Under Section 79 of the IT Act, 2000, social media platforms get legal protection called Safe Harbour. Platforms are not responsible for what is said on stage, but can be held responsible if they have clear knowledge of illegal content and still don't remove it in time.
The decision came after two publishing companies in Munich filed a case, alleging that Google's AI Overviews, the short AI-written summaries you see at the top of search results, were showing false claims about them. The Regional Court of Munich has now issued a temporary injunction in the matter.
Google was also ordered to pay 80% of the legal costs, while the publishers will cover the remaining share.
Publishers challenge AI overviews
The publishers claimed that Google's AI summaries were linking them to scams, subscription traps and other shady business practices. The companies said the claims were false and damaging to their reputation.
The court said that Google's AI Overviews cannot be treated like normal search results and may make the company directly responsible for false information.
AI overviews are not traditional search results
The court further said traditional Google Search only acts as an intermediary, showing links to third-party websites. But AI Overviews are different because they combine information and present it in Google’s own words. This gives users full answers without needing to visit source websites.
"The AI-generated summaries create independent, new, and substantive statements and therefore amount to Google’s own content rather than just third-party information," the court ruled. Judges said Google has full control over how the AI works, including the algorithms behind it. The court said Google is effectively the author and publisher of the AI-generated text.
What court found
The court found that the AI Overview made claims that are not even made in the search results. It also discovered that Google's AI Overview had mixed up information from different sources and, in some cases, created new and incorrect connections that didn't appear in any of the original websites.
Why Google is answerable
Google may not be protected under normal rules that shield online platforms as 'host providers' under the EU Digital Services Act, which usually limits liability for third-party content and requires only notice-and-removal action.
During the court hearing, Google argued that users should not fully trust AI summaries and should always check the linked source websites to verify the information. The company said people already understand that AI-generated content can be wrong, and therefore they should not rely on it blindly.
The court said AI Overviews function like standalone statements, similar to newspaper headlines or teaser lines that can still be legally challenged if they are misleading. "If AI Overviews are treated as publisher speech rather than intermediary indexing, every factual error becomes a potential liability event for Google," the court said.
The judges also said that even if users could check the sources later, that does not remove Google’s responsibility for what the AI already said. The court also warned that if AI answers are treated as "not fully reliable," then the feature itself would lose its value.
One analysis cited in the case suggested Google’s AI Overviews are about 91% accurate, but more than half of even correct answers do not clearly show supporting sources.
Earlier findings
The court also looked at earlier rulings by Germany’s Federal Court of Justice, which had given traditional search engines like Google some protection from liability.
Those earlier decisions said search engines are only ‘indirectly responsible’ because they help users find content from other websites. Courts also said search engines should not be forced to manually check every result, as that would make search services difficult to operate.
But the Munich court said this rule does not apply to AI Overviews.
It explained that a normal search engine only shows links to outside websites. However, AI Overviews read information from different sources, combine it, and generate new sentences in Google’s own voice which the court said they are creating new statements.
"Google can check those statements at least by comparing the underlying third-party websites with its own statements based on them," the court ruled.
The ruling also warned that this issue is not limited to Google. Other AI answer tools, including systems from OpenAI, Microsoft’s Bing, and Perplexity, could face similar liability questions if they generate answers in their own voice.
How social media liability works in India
In India, social media platforms are generally not responsible for user-generated content because they are treated as intermediaries, not publishers. If someone posts illegal content, such as copyrighted material or harmful content, the platform must act once it becomes aware of it.
The Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021, set strict rules for social media companies. These rules state that platforms must act quickly when illegal content is reported, follow proper monitoring and complaint systems and remove unlawful content when required.
Under Section 79 of the IT Act, 2000, social media platforms get legal protection called Safe Harbour. Platforms are not responsible for what is said on stage, but can be held responsible if they have clear knowledge of illegal content and still don't remove it in time.
(Edited by : Sudarsanan Mani)
Source link
