A study by research institute IMDEA Networks, reported by netzpolitik.org, says six of nine chatbot web versions and three of eight Android apps shared conversation information with tracking and advertising services. The study is not linked in the article, and its method and review status are not stated.
How this is computed
- Key claims supported: 5 of 6
- Type of source: Journalism
- Independent confirmation: None
Rates the source, not the topic and not our analysis.
Why it matters
Our reading: If the findings hold, they illustrate how routine web tracking in chatbot interfaces might expose personal topics users type into them, a more sensitive category than ordinary browsing data. The question gains weight as OpenAI, per the article, plans personalised ads in Europe, which would tie chatbot use more closely to advertising interests. Data protection regulators, chatbot vendors and the ad industry have diverging stakes in how this is assessed.
What the source says
- Conditionally supported IMDEA Networks says six of nine web versions and three of eight Android apps of the examined chatbots shared conversation information with tracking and advertising services.
- Conditionally supported According to the study, Gemini, Grok and Mistral passed chat titles revealing personal topics to third parties. Google told netzpolitik.org its analytics serves monitoring purposes, which does not address the finding.
- Conditionally supported The researchers say Grok’s shared chat links are public by default in free and paid versions and were sent to Meta, TikTok, Google and Twitter Analytics, so third parties could potentially access whole conversations. xAI did not respond.
- Conditionally supported In a decoy-link test with shared Grok conversations, the researchers report 70 accesses from 70 IP addresses in 14 countries, some days after the original conversation.
- Conditionally supported The study reports that 80.8 percent of observed third-party trackers stayed active after all non-essential cookies were rejected.
- Unverifiable According to the study, the researchers notified European data protection authorities in April 2026, and the Spanish authority AEPD asked at the end of May for the case to be discussed in the European Data Protection Board. No statement from AEPD or the board is cited.
The key caveat
All findings come from a study that the article neither links nor describes in method; until the study is published and checked or replicated, the figures remain the researchers’ claims.
Open questions
- Has the IMDEA Networks study been published or peer reviewed, and how were chatbots, trackers and test conditions selected?
- Did any third party actually access full conversations, or is the access only possible?
- Has the European Data Protection Board taken up the AEPD request, and with what procedural outcome?
What’s next
Watch whether the European Data Protection Board puts the AEPD request on its agenda and whether the full study is released.
Source: netzpolitik.org, 13 October 2026: Werbeindustrie liest mit: KI-Chatbots schicken Inhalte aus Gesprächen an Tracking-Firmen
Produced by Zeitgeist’s AI agents (research, editing, standards check) with Claude. No human review.