OpenAI is rolling out ChatGPT personal finance to Pro subscribers in the US, in preview on web and iOS, included in the existing $20/month Pro plan with no separate add-on. The integration uses Plaid to connect to "over 12,000 financial institutions" โ named banks and brokers include Schwab, Fidelity, Chase, Robinhood, American Express, and Capital One, with Intuit support planned later for tax impact analysis. The product is a dashboard for portfolio performance, spending, subscriptions, and upcoming payments, plus natural-language financial questions like "help me build a plan to be ready to buy a house in five years." The Hiro personal finance startup team OpenAI acquired in April is the engineering line behind the launch. Disconnect path is Settings โ Apps โ Finances; synced data is reported removed within 30 days.
The thing that needs to be said about this launch, that the launch coverage does not say, is the context. May 13, two days ago, a class action was filed in California federal court alleging that OpenAI piped user chat queries, emails, and user IDs to Meta and Google through the Meta Pixel and Google Analytics tracking tools without obtaining the informed consent California's wiretap statute requires. The complaint cites the California Invasion of Privacy Act and the federal Electronic Communications Privacy Act. The lawsuit is at the allegation stage and nothing is adjudicated. But the technical reading of the suit is that the question being litigated is whether actual chat content โ what users typed โ transited the trackers, or only page-view metadata did. The first reading is the wiretap violation. The second is standard analytics.
Connecting a bank account to a chat service expands the surface of that question dramatically. Plaid intermediation does not change the question; it concentrates it. The chat content that mentions a Schwab balance, a Fidelity transfer, or a Chase Sapphire charge sits inside the same chat infrastructure that the lawsuit alleges was leaking content to third-party trackers. OpenAI's privacy policy disclosures and the Plaid data-handling terms are separate from each other, and the cross-product question โ what content from a finance-connected ChatGPT conversation flows into which ad-tech pixel, if any โ is not addressed in the launch announcement. Builders shipping consumer AI products with financial data have a working example now of how not to time a product launch relative to a wiretap-statute complaint.
The practical builder takeaway is two parts. First, on the consumer side: if you connect a bank account to ChatGPT today, your typed prompts about that bank account are subject to whatever data-handling regime OpenAI's product-side trackers actually implement โ separate from whatever Plaid is doing on its side of the integration. The 30-day disconnect-and-delete path is real but it is a backward-looking remediation, not a forward-looking guarantee about content flow during the connected period. Second, on the builder side: if you ship consumer AI that touches financial PII, the question to answer up front, in your own privacy policy, is which third-party trackers your chat surface loads and what payload they receive โ server-side tagging that strips chat content before it reaches Meta or Google is one architectural answer; a flat ban on marketing pixels on signed-in chat surfaces is another. The OpenAI launch is the question put loudly. The Plaid integration is the amplifier.