Someone poking around Claude’s iOS app found something that isn’t supposed to be there yet. Screenshots published by TestingCatalog on September 14, 2026 show a new “Money” entry sitting in the app’s sidebar, right below Chats and above Code, Artifacts, Dispatch, and Cowork. Tap it, and an onboarding screen reads: “Understand your money with Claude.” Below that, a prompt to link your bank accounts and ask Claude about spending, plans, and more.

Anthropic hasn’t said a word about it. No blog post, no press release, no confirmation it exists. Just interface strings sitting in a shipped app build, discovered by people who go looking for that sort of thing.

I’ve written about this exact scenario before — ChatGPT’s Finances feature launched for real back in May, and it’s $200/month, and it’s genuinely useful, and it’s also a live product you can evaluate today. Claude Money is none of that yet. It’s a leak. Which means the honest thing to do here is separate what’s actually known from what everyone — including me — is speculating about.

What We Know vs. What We Don’t

Discovered	Sept. 14, 2026, via unreleased iOS interface strings
What it shows	A “Money” tab, bank-linking onboarding, prompts to ask Claude about spending and plans
Data provider	Not confirmed (Plaid is the presumed candidate — it powers most of this category)
Pricing	Not confirmed
Launch date	Not confirmed
Supported countries	Not confirmed; US-first is the reasonable guess given banking infrastructure
Official Anthropic statement	None as of publication
Closest live comparison	ChatGPT Finances — Plaid-based, $200/month Pro-only, launched May 15, 2026

What the Screenshots Actually Show

Not much, by design — this is unreleased UI, not a product page. But the pieces that did surface are specific enough to draw real conclusions from.

The “Money” section sits as a first-class tab, not buried in settings. That placement matters. It tells you Anthropic is treating this as core functionality, on par with Code and Artifacts, not a side experiment. The onboarding copy — “link your bank accounts and ask Claude about spending, plans, and more” — is the same basic pitch OpenAI made in May: stop uploading PDF statements, connect once, ask questions forever.

That’s the real shift, and it’s worth sitting with for a second. A statement upload is a one-time snapshot. A linked account is persistent access — Claude (or ChatGPT, or whatever comes next) can see new transactions as they post, indefinitely, until you disconnect it. That’s a fundamentally different trust relationship than “here’s a PDF, tell me about it.”

What Would Claude Money Actually Let You Do?

Based on the interface language and how the equivalent ChatGPT feature already works, a linked-account Claude would plausibly handle:

1. Spending analysis — category breakdowns and trend questions answered in plain language against your real transactions
2. Recurring-payment detection — surfacing subscriptions and repeat charges you’ve forgotten about
3. Balance summaries — a snapshot across connected accounts without opening each bank’s app
4. Budgeting Q&A — conversational questions like “can I afford this” or “how does this month compare to last,” answered with your actual numbers behind them

None of that is confirmed Anthropic functionality. It’s the logical feature set implied by “link your bank accounts and ask about spending, plans, and more,” and it’s exactly what ChatGPT’s version already does. Treat it as an educated guess, not a spec sheet.

Claude Money vs. ChatGPT Finances: What’s Actually Comparable

Here’s the thing about comparing a leak to a live product: the comparison is lopsided by definition. But the shape of it is still useful.

	Claude Money	ChatGPT Finances
Status	Unreleased, unconfirmed	Live since May 15, 2026
Data provider	Unconfirmed	Plaid
Institutions supported	Unknown	12,000+ US institutions
Price	Unknown	$200/month (ChatGPT Pro only)
Access type	Unknown	Read-only
Data policy for financial info	Unknown	Not specifically disclosed beyond general policy

That last row is the one to notice. When ChatGPT’s feature launched, I flagged that OpenAI hadn’t published financial-data-specific handling policies — retention, training use, breach protocols specific to bank data. Four months later, that gap is still the norm for this entire product category, not an OpenAI-specific failure. If Anthropic ships Claude Money without a dedicated financial-data policy, it won’t be an outlier. It’ll be doing exactly what the first mover did.

Anthropic does already have financial-industry plumbing in place through its Intuit partnership, which connects TurboTax, Credit Karma, and QuickBooks data into Claude via an opt-in MCP integration. That’s a different mechanism — it’s Intuit’s data flowing into a conversation, not a direct-to-bank Plaid-style link — but it tells you Anthropic has already built the muscle for “your financial data, inside a Claude chat.” Claude Money would just be the direct-connection version of a pattern the company has already shipped once.

What Anthropic Hasn’t Said (And Why It Matters)

Every meaningful detail is still missing. Not confirmed: the data provider, pricing, which countries or institutions would be supported, whether the connection is read-only, what happens to your transaction data if you cancel, and whether it trains future models. Anthropic hasn’t announced a launch date — this could ship in weeks, or it could sit as dead code in a test build for months. Screenshots of a sidebar tab aren’t a roadmap.

BleepingComputer’s reporting frames it plainly: Anthropic appears to be catching up to a category OpenAI already opened. That’s a reasonable read. It doesn’t tell you anything about how carefully the feature will be built, and building the plumbing to connect a bank account is the easy part — Plaid and its competitors have made that a solved problem, as Plaid’s own $8 billion valuation makes clear. The hard part is everything downstream of the connection: what the AI does with the data, how long it’s kept, and what happens when something goes wrong.

The Security Case Against Linking Any AI to Your Bank

This is where I’d slow down regardless of which company’s logo is on the app.

Every account you connect to anything is another point of failure. That’s not new — it’s the same math behind every Plaid-powered budgeting app you already trust. What’s different with a general-purpose AI assistant is the scope of what it can see and reason about once it’s connected, and the fact that the same account you’d use for coding help or drafting an email is now also the account that can see your balance.

Cybernews reported that security researcher Rachel Tobac, CEO of SocialProof Security, has already warned against linking bank accounts to AI tools generally, describing the risk of an AI account becoming a single point of compromise for attackers looking to drain connected accounts. Her recommendation — strong, unique passwords plus real multi-factor authentication on the AI account itself — applies whether we’re talking about ChatGPT, Claude, or whatever ships next. If someone phishes your Anthropic login and Money is live and connected, they’re not just reading your chat history anymore.

There’s a second layer worth naming: transaction data is more revealing than people assume. Recurring charges can point to health conditions, religious giving, a relationship status, or a job loss before you’ve told anyone. An AI with persistent access to that data, plus the context of everything else you talk to it about, has a more complete picture of your life than a bank statement alone would suggest. None of that means don’t ever connect an account — it means know what you’re actually handing over.

Should You Be Worried Right Now?

No — because there’s nothing live to connect to. If you’re on Claude’s iOS app today, you can’t link a bank account through it, and nobody’s data is at risk from a feature that doesn’t exist for users yet. The leak is a preview, not a launch.

What’s actually worth doing:

• Don’t panic-disconnect anything. Nothing about your existing Plaid-linked apps — Monarch, Copilot, YNAB, whatever you use — changes because of this leak.
• Watch for an actual announcement, not screenshots, before deciding whether Claude Money is worth using. Pricing and data policy are the two facts that determine whether this is worth your time, and neither exists yet.
• If security and privacy are a priority for you — and per our editorial position on this, they should be — wait for a published, financial-data-specific policy before connecting anything. General AI terms of service aren’t written with bank-transaction sensitivity in mind, and that gap has already tripped up the first company to ship this kind of feature.
• Compare the eventual price to what you’re already paying. ChatGPT’s version costs more per month than most dedicated budgeting apps cost per year. There’s no reason to assume Claude Money launches cheaper — general-purpose AI subscriptions aren’t priced like budgeting apps, they’re priced like AI subscriptions.

The Bottom Line

Claude Money is real in the sense that the code exists somewhere in Anthropic’s app. It is not real in the sense that matters to your actual bank account: there’s no price, no data policy, no confirmed provider, no launch date, and no way to use it. Everything past that is informed speculation built on how the ChatGPT equivalent works — useful for setting expectations, not a substitute for Anthropic’s own disclosure.

The pattern underneath this leak is the one to actually track. First OpenAI, now (apparently) Anthropic — general-purpose AI assistants are converging on persistent bank-account access as a default feature, not a specialty product. That’s a bigger shift than any single company’s rollout, and it’s coming whether or not you personally connect an account. Worth having an opinion on before it’s your only option.

---

Reporting based on screenshots published by TestingCatalog on September 14, 2026, and subsequent coverage from BleepingComputer and Cybernews. Anthropic has not confirmed Claude Money’s existence, pricing, data provider, or launch timeline as of publication. Details here will be updated if and when Anthropic makes an official statement.