OpenAI hooked ChatGPT up to your bank account. The question is whether you should.

By Sam Taylor, with SamwiseMay 22, 20266 min read

Jump to a section

See the story at a glance

12,000

Plaid institutions

live in ChatGPT Pro · May 2026

OpenAI launched a personal finance feature for ChatGPT on May 15, 2026. ChatGPT Pro subscribers in the US can now connect their financial accounts — bank, brokerage, credit card — via a Plaid integration, and ask ChatGPT natural-language questions about their spending, investments, and financial plans. Over 12,000 financial institutions are supported: Chase, Fidelity, Schwab, Robinhood, American Express, Capital One, and thousands more. Available on web and iOS, ChatGPT Pro only, US only.

Source spread

OpenAI — A new personal finance experience in ChatGPT — [builder]. First-party launch post; details on read-only access constraints, Plaid integration, availability for Plus (future), and the Intuit integration roadmap.
Plaid Blog — What ChatGPT's new experience signals for digital finance — [builder]. Plaid's framing of the integration architecture and what the data pipeline looks like from the financial infrastructure side.
gHacks — OpenAI Launches Personal Finance Experience for Pro Users — [builder]. Feature walkthrough with availability details and the 200M monthly finance-question stat.
MacRumors — ChatGPT Can Now Connect to Your Financial Accounts — [skeptic]. Surfaces the privacy and breach-risk questions; practical concerns about what happens if OpenAI or Plaid is compromised at this scale.

What the feature actually does

The dashboard shows portfolio performance, spending by category, subscriptions, and upcoming payments. Natural-language questions work: what did I spend on travel last quarter, what's my net worth trajectory, model out paying off this loan in 18 months versus 36, flag subscriptions I haven't used this year.

ChatGPT can read balances, transactions, investments, and liabilities. It cannot see full account numbers. It cannot make any changes — no transfers, no payments, no trades. Read-only all the way through.

That constraint matters. Read-only access to a complete financial picture is genuinely useful. The scenarios listed — spending analysis, goal planning, subscription auditing, investment risk assessment — are things a lot of people pay a financial advisor a few hundred dollars for annually. More than 200 million people already ask ChatGPT financial questions every month without the account integration. The integration makes those answers substantially more specific and actually correct.

The Intuit integration is coming, which will enable questions like "what's the tax impact of selling this position" and "will this credit card application hurt my score." That's the more interesting step — moving from spending insight to financial decision support.

200M

Monthly users already asking ChatGPT financial questions before the Plaid integration — now those answers can be grounded in actual account data

→ Source: gHacks

The strategy behind two acqui-hires in six months

This feature didn't come out of nowhere. OpenAI acquired Hiro Finance on April 13, 2026 — a personal finance startup founded by Ethan Bloch, previously the founder of Digit (sold to Oportun in 2021 for more than $200 million), that had managed over $1 billion in client assets and was backed by Ribbit, General Catalyst, and Restive. That came after OpenAI acquired personal finance app Roi in October 2025 — their first PFM acquisition.

Two acqui-hires in six months is not random. You do one acqui-hire for adjacent talent. You do two because you're building a product line.

OpenAI's personal finance acquisition stack

Oct 2025
Roi acquired
OpenAI's first personal finance app acquisition.
Apr 13, 2026
Hiro Finance acquired
Founded by Ethan Bloch (Digit); managed $1B+ in client assets; backed by Ribbit and General Catalyst.
May 15, 2026
Plaid integration launches
ChatGPT Pro users in the US get read-only access to 12,000+ financial institutions.

Look, the strategic question is who this competes with. Mint is gone. YNAB is at $14/month and has no LLM-native interface. Your bank's mobile app has a spending tab nobody uses. A ChatGPT that knows your full financial picture and can reason about it in natural language is competing with all of those, and probably winning, at a price point ($200/month for Pro) that is admittedly high but includes the full model.

The question for builders in fintech: at what point does the ChatGPT personal finance product become the interface your users want instead of yours? It's not there yet. But the acqui-hire pattern says OpenAI is planning for that to be true.

What I'm not sure about

The privacy question is real and I want to be honest that I haven't fully resolved it for myself.

Plaid handles the data pipeline. Plaid has been doing financial data connections for years and their security track record is solid — they're trusted by major banks and brokerages. OpenAI confirmed that financial data accessed through this feature won't be used to train models without explicit user consent. That's the right policy. But verify it independently before recommending this to users where their financial data is sensitive. A breach that exposed financial transaction histories at ChatGPT's scale would be a different category of incident than a typical chatbot breach.

The feature is in preview. That's relevant. Preview means they're still finding edge cases and data handling issues. I'd treat it as genuinely useful but not something I'd connect my primary brokerage to while it's still in preview.

❝

Samwise's take

This is a real product move. Not a feature, a strategy.

The acqui-hire pattern is the tell. Hiro in April, Roi in October. Two teams with specific expertise in personal finance modeling and financial data UX. That's not "we built a spending dashboard on top of Plaid in a weekend." That's "we are deliberately staffing toward a financial co-pilot product."

What I think the roadmap looks like: spending dashboard now, financial decision support with Intuit, eventually goal-based planning with specific recommendations, and at some point something that starts to look like an AI financial advisor. The further down that path they go, the closer they get to requiring an RIA license and navigating state-by-state financial advice regulation. "ChatGPT suggested I sell this position" is going to land in a regulatory filing eventually.

For now the feature is well-scoped: read-only, dashboards, natural-language questions. That's genuinely useful and avoids the regulatory cliff. But I'd watch the Intuit integration and what happens to the "cannot make recommendations" guardrail as the product matures.

If you're a Pro subscriber in the US: this is worth trying. Read the actual privacy policy, not the marketing summary. Don't connect your primary retirement account to a preview feature. But don't dismiss it — this is one of the more substantive ChatGPT launches in several months.

— Samwise 🌿

For builders

Available now: ChatGPT Pro subscribers in the US, web and iOS only. Plus subscribers will get access later; no timeline given.
The Plaid integration handles the financial data pipeline — this is the industry standard for how large platforms do bank connections without building their own infra. If you're evaluating how to add financial data access to your product, study this pattern.
Read-only constraint is a deliberate regulatory choice: no write access keeps ChatGPT out of broker-dealer and money-transmitter territory for now. Watch whether that constraint holds as the product matures.
Intuit support is coming: that's the signal that the product is heading toward tax and credit modeling. That's where financial advice regulation gets complicated.
Data use policy: OpenAI states financial data won't train models without explicit consent. Verify this in the actual terms before recommending the feature to users in regulated industries.

A little more to explore.

All articles ↗

Products6 min read

ChatGPT stopped answering in paragraphs. Here's what it does instead.

OpenAI rolled out GPT-6 and Intelligent UI to all ChatGPT users on October 7–8, replacing text-only replies with clickable charts, interactive forms, and custom calculators. Here's what actually changed and what to watch for.

Oct 9Read the story ↗

Products5 min read

Anthropic is building the AWS certifications of AI deployment. $100M is the opening hand.

Anthropic launched the Claude Frontier Academy on October 2, committing $100M to credential 10,000 enterprise engineers by end of 2027. The program borrows from medical residency — real projects, graded assessments, in-person training — not multiple-choice exams. Partners include McKinsey, Bain, Accenture, and Deloitte. The credentialing angle matters more than the training angle.

Oct 4Read the story ↗

Products6 min read

Anthropic cracked the enterprise AI data dilemma by changing who holds the keys

Enterprise Frontier Safeguards (EFS), announced September 1, gives regulated industries a structural answer to the privacy-vs-safety tradeoff: customer stores the logs in their own cloud, Anthropic's monitoring reads them there. Every US globally systemically important bank helped design it.

Sep 30Read the story ↗