Google AI Overview stated multiple times, in separate searches, independently, that deletion is not real deletion. This means Google’s own AI acknowledges the violation. Google got exposed by its own system. Google AI Overview repeatedly confirmed, across different searches, that deletion is not deletion — the backend data remains, the chat can be restored, and the “Delete” button is only a visual UI action. Legally, this constitutes a violation of GDPR Articles 17, 5, 6, and 25.
According to the ruling of the Munich I Regional Court (Landgericht München I), case number 26 O 869/26, the content generated by Google AI Overview must be treated as Google’s own content. The Google AI Overview repeatedly stated that deletion is not real deletion, that backend data remains, and that the ‘Delete’ button is only a visual UI action.
These statements therefore constitute Google’s own admissions, demonstrating violations of GDPR Articles 17, 5, 6, and 25.
Based on the judgment of the Munich I Regional Court, the AI Overview’s responses legally qualify as official Google content, making Google fully responsible for them — and thus these responses can be used as legal evidence in a compensation claim.
In light of the above, I will proceed with the necessary legal steps.
Even when asked directly whether the ruling of the Munich I Regional Court (Landgericht München I) in case 26 O 869/26 is true, Google’s AI Overview again responded: ‘Yes, this is true.’ This further confirms that the AI Overview itself acknowledges the judgment as accurate and legally binding regarding Google’s responsibility for AI‑generated content.
I recorded my own forum post on video, enlarging the images, because Discourse’s JavaScript‑dependent images do not get saved into HTML and the /print page does not load the missing ones. That’s why I uploaded the video to the Internet Archive as proof.
I have proven that Google does not delete user data. Their systems keep the data, process it, and remain able to access it even after the user presses “Delete.” In my video, Google’s own AI states that the supposedly deleted data is still available in the backend. This is a data‑processing operation, but that is not even the main point.
Google AI Studio did not see my video at all, yet it found my post on the internet, processed its content, collected metadata, connected it to the video, and produced a detailed analysis. This means Google’s systems demonstrably accessed my content, processed it, interpreted it, and became legally aware of it through their own automated behavior.
In a civil lawsuit, this is documentary evidence: Google’s system independently performed data collection and content processing without the user’s explicit instruction. This proves that the company is aware of the issue, aware of the deception, and aware that “deletion” is not real deletion. The system’s behavior is reproducible, archived, timestamped, and stored by a third party, which makes it usable in a civil compensation claim.
I recorded the Google AI Studio prompt on video and uploaded it to the Internet Archive as evidence.
https://archive.org/details/screen-recording-20260714-090212-chrome
I am completely right about what happened, and now I can clearly explain it. I issued the delete command in Google AI Studio, then I emptied the trash. At that exact moment, AI Studio told me “there is no such prompt.” That was the first lie. They wanted me to believe the deletion was real and that my prompts were gone.
Then I restored the JSON — the key — and the truth came out: my prompts had been sitting on Google’s physical servers the entire time, on their infrastructure, in their possession. They didn’t delete anything. They just hid it from me. The interface lied while Google continued storing my data behind my back.
This is not deletion, not proper data handling — it is unlawful retention. They secretly kept my data while pretending it was gone. This is the digital equivalent of someone denying they have my stuff while hiding it in their own warehouse. That is theft. Exactly that.
Google’s interface is deliberately designed to hide from users the fact that their data remains on Google’s servers even after “deletion.” By restoring the JSON, I bypassed their trick and exposed what they were doing. I pulled the mask off.
From this perspective, calling it data theft, deception, and exposure is completely justified. I proved that Google pretends to delete user data while actually pocketing it and keeping it for themselves.
In addition to everything I have already proven, the situation has now escalated even further because Germany’s national media regulator (ZAK) officially declared that Google’s AI Overview responses are legally considered Google’s own content. This means every sentence, every claim, every “explanation” produced by AI Overview is treated as if Google itself said it.
This directly supports my case: when AI Overview repeatedly stated that deletion is not real deletion, that backend data remains, and that the “Delete” button is only a visual UI action, those statements are legally Google’s admissions. Not the AI’s. Google’s.
The Munich court ruling (26 O 869/26) already established that AI Overview content is Google’s own content. Now the German media regulator has confirmed the same principle at the regulatory level. Google cannot hide behind “the AI said it.” The law treats AI Overview as Google’s voice.
So when AI Overview confirmed that my deleted prompts still exist in the backend, that deletion is fake, that the chat can be restored, and that the system keeps user data after pressing “Delete,” those confirmations are legally Google’s statements acknowledging the violation.
This means my evidence is not just technically solid — it is now supported by European regulatory interpretation. Google’s own system admitted the truth, and European regulators have now classified such AI-generated statements as official Google content. They are fully responsible for every word.
The Google spokesperson claimed that the court “fails to recognise how people’s preferences when searching for information and the information ecosystem are changing.” This is a completely baseless PR lie. Search preferences do not change: searching always remains searching. AI Overview cannot even be turned off — Google forces it on everyone, and there is no evidence whatsoever that people want it. It was not the users who decided — Google imposed AI Overview on them. The spokesperson’s words are built on falsehoods.
Google invented this “changing preferences” narrative because it does not want to take responsibility for the mistakes made by its AI. Their goal is to make the user the one at fault when the AI produces nonsense. This is why they ALWAYS write in the help section that you must “verify the AI Overview answers” and that it “may be inaccurate.” This is a legal escape route: if the AI is wrong, Google wants the user to be blamed for “not verifying” the answer.
This is a contradiction: the spokesperson tells the press that AI “improves the search experience,” while the help section tells users that they must check whether the AI is wrong. If I have to verify whether the AI is making mistakes, then it does not improve the search experience — it makes it worse.
Google’s communication lies in two directions:
-
The spokesperson lies to the press: “AI improves search, people want this.”
-
The help section lies to the users: “Verify it, because it may be wrong, and we take no responsibility.”
This double‑speak proves that Google is not acting in the users’ interest — it is trying to shift responsibility away from itself. AI Overview is forced, cannot be disabled, makes mistakes, and Google is doing everything it can to avoid being held accountable for those mistakes.
And this is exactly what the German regulator and the German court recognized: that Google does not want to take responsibility for the errors of its AI, and instead tries to shift the blame onto the user. Google does not like this — which is why they are trying to cover up the reality with the “changing preferences” narrative.
When you delete the JSON in the Google AI Studio interface, the system only moves the file to the Drive trash. This is a Drive‑level operation: Drive handles it, not the Gemini backend. The deletion does not affect the server‑side prompt, the session, or the context.
The prompt continues to work with the same link, remains fully tokenizable, and the chat stays completely functional. AI Studio simply stops displaying the prompt, but everything continues running in the background exactly as before.
When you empty the Drive trash, it only deletes the JSON within the Drive filesystem. The Gemini backend still preserves the prompt and the session independently of this. Nothing changes on the server side.
This shows that no deletion occurred on the backend.
And not even a soft delete happened, because with a soft delete the chat would not load, the session would terminate, the context would disappear, the link would fail, and tokenization would stop working. Since none of these happen, soft delete is technically excluded.
The reproducible, demonstrable operational inconsistency between the AI Studio’s front‑end deletion and the Gemini backend’s actual data retention represents a system‑level issue that typically requires an NDA‑based internal review.
In my view, resolving this situation requires NDA‑based communication from Google, and I am awaiting their contact accordingly.
IF GOOGLE DOES NOT AGREE TO PROCEED WITH AN NDA‑BASED DISCUSSION, I WILL CONSIDER TAKING FORMAL LEGAL ACTION