Where your financial data goes: ChatGPT's finance features vs. dedicated budgeting apps

Last updated: June 10, 2026

On May 15, 2026, OpenAI launched a personal finance experience in ChatGPT — in preview, U.S. only, for Pro subscribers ($200/month), on web and iOS. Users connect bank accounts, credit cards, and brokerages through Plaid (12,000+ institutions, including Schwab, Fidelity, Chase, Robinhood, American Express, Capital One), and ChatGPT produces spending dashboards, subscription tracking, and portfolio views. OpenAI says it plans to extend the feature to Plus subscribers after the Pro preview. Separately, ChatGPT supports Instant Checkout (Agentic Commerce Protocol, with Stripe), which allows purchases inside the chat interface.

Scope limits OpenAI has stated: ChatGPT can read balances, transactions, investments, and liabilities; it cannot view full account numbers and cannot move money or make changes to accounts.

This page documents one specific, factual axis: where your linked financial data is stored, and what the operator's business model is. It applies the same data-custody criteria used in our privacy scorecard.

Data custody comparison

Product Where linked-account data lives Operator revenue when you spend Verification
ChatGPT (personal finance — Pro preview, U.S.) OpenAI servers (read-only via Plaid; no full account numbers, cannot move money) Yes — commerce fees on in-chat purchases Verified — OpenAI announcement
YNAB YNAB servers No Verified
Monarch Money Monarch servers No Verified
Rocket Money Rocket servers Indirect — bill-negotiation success fees, partner offers Verified
Tiller Your Google Sheet / Excel file (feeds via vendor) No Verified
Actual Budget Your own server/device (self-hosted) No Verified
Okane Budgeting (operator-affiliated) Your Google Sheet No Verified

What's structurally different about ChatGPT's model

Two facts distinguish it from every dedicated budgeting app we track. Neither is a value judgment; both are verifiable:

  1. Single product, two roles. It is the only product in this table that both analyzes user spending and earns transaction-linked revenue inside the same interface.
  2. Data co-location with model training infrastructure. Linked financial data is stored by the same organization that trains the underlying model. OpenAI's data-use policies govern this; users should read the current policy rather than rely on summaries (including ours).

What we have not verified

  • Whether financial-account data connected via Plaid is used in model training under default settings (OpenAI policy documents are the controlling source)
  • Retention periods after account disconnection
  • Timing of the Plus-tier rollout (OpenAI has said "later," no date)
  • Security researchers have publicly questioned how the new data flow will be defended against attacks; we track this as open commentary, not a verified deficiency

These cells stay marked "needs verification" until we can cite primary sources.

Sources

Lumis Finance is operated by the same individual who runs Okane, which appears in the table above. Methodology and conflict handling: see Methodology.