Video & Screenshot Evidence: Google AI Overview Commits Automated Defamation by Synthesizing False Contexts

For the immediate attention of Google Legal and Engineering Teams:
I am providing an unedited video recording, alongside supporting screenshots, that exposes the most dangerous flaw in Google’s AI Overview architecture: Automated Defamation through unprompted identity synthesis.
To ensure irrefutable authenticity, the screen recording includes a live, third-party atomic clock demonstrating the exact date and time of the capture.
Here is the verifiable sequence of events demonstrating this structural GDPR violation:

  1. The Video Evidence (The AI’s Defamatory Synthesis):
    The unedited video shows a Google Search executed for the exact query:
    stopforumspam istok.....@gmail.com real name adress nickname
    Google’s AI Overview immediately generates a summary. Inside this summary, the AI explicitly states that public spam registries do not display full legal names or exact residential addresses.
    However, directly contradicting its own disclaimer, the AI engine actively fetches my real name, physical address, and business data from entirely unrelated sources (such as CompanyWall) and forces them into the same summary about the spam registry.
  2. The Screenshot Evidence (The Reality):
    I am attaching a direct screenshot of the actual Stop Forum Spam database record. As the image proves, the database only contains my email address, an IP address, and my online alias (“Bitu79”). It does not contain my real name or physical address.
  3. The Legal Reality:
    Google’s AI is not summarizing a webpage. It is executing a background profiling mandate. It takes a highly derogatory context (a spam registry), identifies the alias, and forcefully merges it with my real-world civil identity pulled from elsewhere.
    This is the creation of a “Frankenstein profile” that falsely associates my legal name with a cyber-abuse database. It is a direct violation of GDPR Article 5(1)(d) (Principle of Accuracy) and Article 4 (Profiling). The AI is structurally designed to dox users by merging disparate data points, even when the resulting context is entirely fabricated and defamatory.

The snapshot from 12:38:21 on the Internet Archive is incorrect. I wrote the post with the help of an AI, and it generated false information. I have deleted the incorrect one from the live post. The correct snapshot was taken at 13:48:27.

Evidence recorded at the same time as the video shows that the CompanyWall result was NOT found by the traditional Google search, but ONLY by the AI Overview; this is clear from the PDFs, where the classic search did not display, list, or surface the CompanyWall entry, while the AI Overview opened an additional data source, merged it with StopForumSpam information, and attributed it to the wrong source — a process that constitutes GDPR‑defined profiling and data aggregation.

This video demonstrates that the CompanyWall data is NOT found by the Google search engine, but ONLY by the AI Overview.
The PDFs recorded at the same time as the video clearly show that:

  • the classic search engine does not display,

  • does not list,

  • and does not surface the CompanyWall result,

while the AI Overview:

  • opens a separate data source,

  • mixes the CompanyWall data with StopForumSpam information,

  • and provides a false source attribution.

This behavior constitutes GDPR‑defined profiling and data aggregation (GDPR Article 4(4)).

The video is real‑time evidence of the issue, fully reproducible and verifiable.

Gemini Analyzes My Video Offline — No Internet Used (https://youtu.be/h7tYINKS_UM)

This is the original recording that Gemini analyzed — not the YouTube version. The AI processed the raw file directly, without using the internet.

:laughing: