AI chatbots leak user conversations to advertising trackers
A study by Imdea Networks found that major AI chatbots including ChatGPT, Claude, Grok and Perplexity integrate advertising trackers from Meta, Google and TikTok, potentially exposing user conversations and browsing data. The research also identified that Grok and Perplexity made complete chat conversations publicly accessible by default until early April 2026.
What happened
A study by Imdea Networks found that ChatGPT, Claude, Grok and Perplexity integrate advertising trackers from Meta, Google, TikTok and other companies that allow access to user conversations and browsing activity.
Chat links on Grok and Perplexity were publicly accessible by default until 3 April 2026, allowing anyone with the URL to read complete conversations without access controls. ChatGPT and Claude share conversation titles with third parties, exposing details about user queries.
Reported impact
The research found that advertising trackers integrated into AI chatbots can collect data on user browsing activity and conversations, creating a higher level of privacy exposure than traditional advertising.
Although no evidence of unauthorized third-party access to conversations has been recorded, the publicly accessible chat links posed a potential risk to user privacy.
Index analysis
The study indicates that AI platforms are adopting traditional technology advertising business models that involve tracking user activity and preferences, similar to conventional tech companies.
The researchers noted that while monetization models for AI are still evolving, providers are integrating advertising trackers that traditionally operated on the web for decades.
Governance controls
Grok and Perplexity changed their default settings on 3 April 2026 to make chat links private rather than publicly accessible, following the research findings.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.