OpenAI targeted in CIPA lawsuit over sharing sensitive ChatGPT queries with advertisers

Amargo Couture filed a proposed class action against OpenAI Global, LLC on May 13, 2026, in the US District Court for the Southern District of California. The complaint alleges that OpenAI embedded Meta Pixel and Google Analytics into the ChatGPT.com website, intercepting sensitive user queries and leaking them to Meta and Google for advertising purposes.
- The complaint alleges that when a user types a query into ChatGPT, the website simultaneously sends duplicate data (including the query itself) to Meta and Google alongside advertising cookies and personally identifiable information
- The lawsuit brings claims under the California Invasion of Privacy Act (CIPA) §§ 631 and 632, the federal Electronic Communications Privacy Act (ECPA), and California constitutional privacy protections
- Meta allegedly receives queries linked to unencrypted Facebook IDs via the c_user, fr, and _fbp cookies, allowing it to tie ChatGPT conversations to real-identity Facebook profiles
What is this case about?
The plaintiff alleges that OpenAI's website code included tracking technologies from Meta and Google that operated alongside the ChatGPT interface. When a user submitted a query, the same data was allegedly transmitted to those third parties in parallel.
For Meta, the data reportedly included the query text alongside Facebook identifier cookies containing the user's unencrypted Facebook ID. This would allow Meta to connect a user's private ChatGPT questions to their real Facebook profile.
For Google, the complaint alleges that Google Analytics intercepted IP addresses, device IDs, the _ga cookie, the Secure-3PSID cookie, and hashed login email addresses. The plaintiff claims Google can link this data back to existing user profiles and feed it into Google Ads and other products.
Why does CIPA apply here?
The complaint uses two sections of CIPA.
- Under § 631 (wiretapping), the plaintiff alleges that Meta Pixel and Google Analytics functioned as "machines, instruments, or contrivances" that intercepted the contents of communications in transit
- Under § 632 (eavesdropping), the plaintiff argues that users have a reasonable expectation of privacy when asking a chatbot sensitive medical, financial, or personal questions, and that the tracking codes allowed third parties to eavesdrop on those confidential communications
This is the same legal theory that has been applied to traditional website tracking in cases involving publishers and ecommerce sites.
The difference here is the sensitivity of the data: ChatGPT conversations can include deeply personal medical, legal, and financial queries, which strengthens the claim that users expected privacy.
What does this mean for advertising?
The complaint alleges Meta used the intercepted data to build "Custom Audiences" and "Lookalike Audiences" for targeted ads. Google allegedly integrated it with automated ad bidding and cross-device remarketing tools.
This case extends the growing line of CIPA litigation beyond traditional publishers and ecommerce into AI platforms. Any website that embeds third-party advertising tags on pages where users submit sensitive inputs faces similar exposure.
If a tracking pixel can read form fields, search queries, or chat prompts, that data may be transmitted to the tag vendor's servers regardless of your intent. Teams should audit exactly what data each tag can access, whether it fires before or after consent, and whether any user inputs are included in the payload.
Key takeaways
- Establish robust digital tracking governance to track all personal data elements shared and all third parties receiving personal data
- Continuously audit websites and apps to ensure that user consent is actually honored and no sensitive data is shared
- Run data protection assessments for any processing of personal data for targeted advertising, selling of personal data, or processing of sensitive data
Privado AI's Web Auditor continuously scans websites to verify that no tracking pixels or analytics tags transmit user data before consent is captured, and flags any scripts with access to sensitive input fields.



