A new class-action lawsuit claims OpenAI’s ChatGPT website used tracking tools embedded in the site to share users’ query topics and identifying information with Meta and Google. As more and more people turn to AI chatbots for everything from sensitive questions about personal problems to financial concerns, the case is raising fresh questions about how private those interactions really are.

The complaint, filed on May 13 in the U.S. District Court for the Southern District of California, alleges that OpenAI disclosed query topics and Facebook-linked identifiers to Meta, while Google allegedly received query topics, user IDs, and hashed email addresses through Google Analytics.

The plaintiff, California resident Amargo Couture, says she used ChatGPT in 2025 and 2026 to ask about “sensitive information about her health, finances, and other private information.”

“Despite reasonable expectations of privacy,” the lawsuit claims, OpenAI “disclosed information provided by consumers to Meta Platforms, Inc. … and Google, LLC … by incorporating technology owned by each third party into the code of its website.” The suit brings claims under the federal Electronic Communications Privacy Act, the California Invasion of Privacy Act, and California privacy law.

While the allegations haven’t yet been tested in court, the case comes as AI companies are selling themselves as places where people can go for help with their work, questions, and day-to-day decision-making.

In a statement about bringing ads to its platform, OpenAI says, “People trust ChatGPT for important and personal tasks.” One of the questions this lawsuit raises is whether that trust is misplaced.

Lawsuit Says ChatGPT Shared Query Topics, User IDs & Email Data

The lawsuit claims that ChatGPT query topics were sent to Meta and Google in real time through ordinary website tracking tools. According to the complaint, when a user entered a prompt into ChatGPT, the site generated a browser tab title reflecting the subject of the query.

In one of the examples included in the filing, a user asks, “Who won the Super Bowl in 2005?” The browser tab then displays the title “Super Bowl 2005 Winner.” The screenshots show the phrase appearing in the data allegedly sent through Meta Pixel and Google Analytics.

“When users enter queries into ChatGPT,” the lawsuit says, “Defendant discloses that information to Meta in real time.” The filing makes the same allegation about Google.

The complaint goes on to say that Meta received query-topic information, along with identifiers associated with users’ accounts, including data associated with cookies such as c_user, fr, and _fbp. It also claims that Google received query topics along with user IDs and hashed email addresses when people created or logged in to ChatGPT accounts.

As spelled out in the lawsuit, the plaintiff regularly used ChatGPT in the same browser she used to access Facebook and stayed logged into Google while using the platform. The complaint presents that overlap as part of its theory for how those identifiers could connect chatbot activity to a specific person.

In the filing, there’s nothing that establishes that Meta or Google received every word users typed into ChatGPT. Its clearest allegations involve query topics or titles derived from prompts, as well as identifying data that the complaint says could link those topics to users.

The suit further alleges that Meta and Google used that information “for [their] own purposes, including for targeted advertising.”

Why ChatGPT Privacy Feels Different From Ordinary Website Tracking

Web users have spent decades navigating an internet built around tracking pixels, analytics tags, and behavioral advertising. Even those who aren’t especially tech-savvy have some understanding of the fact that websites monitor clicks, page views, and shopping activity.

However, ChatGPT is different. Its chat-based interface can make it feel more like a private workspace or sounding board than a webpage on the public internet. It invites users to ask questions that they may not be comfortable posting publicly.

Many use it to test out ideas, talk through personal issues, and seek help with problems they haven’t figured out on their own. For many users, it’s become a personal diary of sorts. And many users expect their conversations to remain between them and what feels like a private chatbot.

It’s that expectation that the lawsuit turns on, when it says, “Plaintiff and Class Members expected their communications to be confined to Defendant.” Research supports the idea that users behave differently when interacting with conversational AI. A 2025 paper in Philosophy & Technology argues that “forming a recurring quasi-relationship with an AI agent can elicit deeper disclosure, as the user feels a sense of trust or even empathy toward the system.” The authors say those interactions create new privacy concerns because AI agents can store, analyze, or expose that personal information in ways a human confidant would not.

While AI chatbots like ChatGPT typically have extensive privacy policies, the reality is that most users aren’t reading every line of these disclosures. OpenAI has recently tried to explain some of its marketing-related data practices in plain language.

In an article updated this month, the company says it shares some information with “social networks and large advertising platforms” to promote OpenAI products on third-party sites and apps. That can include whether someone signed up for ChatGPT’s Free tier or visited a page about a specific OpenAI product. OpenAI says it does not share conversations or uploaded content with those marketing partners.

The company also says the identifiers it shares do not reveal a person’s real-life identity. Instead, it describes them as “a string of numbers and letters linked to identifiers such as your browser or device (a cookie ID), your IP address, or a version of your email address or phone number run through a security tool called a ‘hash’ making it into a string of numbers and letters.”

However, the lawsuit raises a separate question: whether, as the complaint alleges, query-topic information from ChatGPT itself entered tracking systems tied to Meta and Google. The possibility of that happening hits differently in the context of a chatbot that people use to discuss illness, debt, legal worries, or family problems than it would for a typical marketing page.

The OpenAI Lawsuit Highlights a Bigger AI Privacy Problem

It’s much too early to know whether this lawsuit will succeed or fail as it moves through the legal system. Still, whatever the ultimate outcome, the issues it raises aren’t going anywhere.

The way chatbots work makes it very easy for users to move more of their personal lives into these conversational interfaces. They can summarize uploaded documents, remember preferences, help out with work, and respond to questions that users once posed to close friends or professional advisers. The more intimate these exchanges become, the more pressure these companies will face to explain what happens to the data users entrust to them.

A 2026 Stanford Institute for Human-Centered AI issue brief warns that foundation models create privacy risks across their life cycle, including “the intimate data that users unwittingly disclose through chatbot interfaces.” It also notes that chatbot interactions can reveal “far more emotional and psychological information” than search histories alone.

The lawsuit against OpenAI asks whether users understood that personal questions entered into a chatbot could allegedly travel alongside identifying data into the same tracking ecosystem that powers much of the commercial web.

As AI assistants become places where people think out loud, companies may find that familiar website disclosures no longer match the expectations their products create.

Originally published by Techopedia on May 18, 2026.