Anthropic tests Claude Money, because chatbots need your bank balance too

Why it matters

Claude Money would move Anthropic from generic financial answers to persistent analysis of real account data, putting privacy, accuracy and bank integrations at the center of its consumer strategy.

Anthropic is testing a personal-finance section inside Claude that would let users link bank accounts and ask the assistant about spending, plans and other money questions, according to screenshots posted by TestingCatalog on X on September 14th.

The unreleased iOS interface places "Money" alongside Chats, Code, Artifacts and other Claude products. A separate onboarding screen reads "Understand your money with Claude" and prompts users to link their bank accounts. TestingCatalog reported that the feature has appeared in the mobile app's interface, though it has yet to surface on Claude's web product.

Anthropic has not announced Claude Money or published details about its availability. The screenshots do not identify the account-connection provider, supported institutions, eligible subscription plans or geographic scope. They also do not establish whether Claude Money would have read-only access or eventually support actions such as transfers and bill payments.

Those details will determine whether Claude Money amounts to a transaction-search interface or a more consequential financial agent. Persistent access to account balances and transaction histories would let Claude answer questions using a household's actual financial position instead of requiring users to upload statements or explain their finances in each conversation.

Co-founders Dario Amodei and Daniela Amodei continue to divide Anthropic's leadership between research strategy and commercial execution. Dario Amodei serves as CEO and directs research, while Daniela Amodei, Anthropic's president and board chair, oversees product, engineering, governance and commercial operations. Claude Money would push their consumer product into one of the most sensitive data categories Anthropic has attempted to handle directly.

Anthropic has been assembling the finance pieces

Claude Money would extend a finance push that has so far centered on professional users and third-party integrations. Anthropic recently released ten finance-agent templates for jobs including pitchbook preparation, financial modeling, ledger reconciliation, statement auditing and know-your-customer reviews. Those products are designed for Claude Cowork, Claude Code and Claude Managed Agents rather than household budgeting.

Anthropic also signed a multiyear partnership with Intuit in February. The agreement called for TurboTax, Credit Karma, QuickBooks and Mailchimp tools to become available inside Anthropic products through Model Context Protocol integrations. Intuit said consumers would be able to connect financial data to Claude for tasks such as estimating tax refunds and reviewing high-interest debt.

A native Money tab would give Anthropic control over the entry point instead of leaving personal finance to independent connector developers. It could also create a recurring reason for consumers to return to Claude: bank balances, subscriptions and spending patterns change continuously, unlike many one-off chatbot queries.

OpenAI got there first

Anthropic would be following an established product pattern. Finances in ChatGPT already allows eligible US Plus and Pro subscribers to connect accounts through Plaid and review spending, bills, subscriptions, net worth and investments on web, iOS and Android.

OpenAI's implementation sets a clear baseline for what Anthropic will need to explain. ChatGPT's finance product is read-only: it cannot move money, pay bills, make trades or alter account settings. OpenAI also tells users that ChatGPT is not a fiduciary, investment adviser, broker-dealer or tax preparer.

The competitive pressure is straightforward. A general-purpose assistant becomes harder to replace when it remembers a user's financial position and can answer questions against live account data. The same integration raises the cost of a mistaken answer. Categorizing a restaurant charge incorrectly is an annoyance; misreading debt, cash flow or investment exposure can affect decisions carrying real losses.

The privacy policy will be part of the product

Anthropic's current connector documentation says connecting a service can grant Claude permission to access and potentially modify data according to the user's underlying account permissions. Anthropic advises users to review each connector's capabilities and connect only services they trust.

For consumer accounts, Anthropic says raw content retrieved from connectors is excluded from model training. Its training policy also notes that information copied directly into a conversation, including details appearing in Claude's responses, may be covered by a user's broader training settings.

Claude Money therefore needs controls built for financial records rather than generic connected apps. Users will need to know where account data is stored, how long it remains after disconnection, which portions can enter chat history or memory, and whether every connection is read-only. The interface found by TestingCatalog answers none of those questions yet.

The screenshots show that Anthropic has advanced beyond a product name: Claude Money has a dedicated navigation slot, an onboarding pitch and a bank-linking prompt. Shipping it will require Anthropic to persuade consumers that an AI assistant can be trusted with a running ledger of their financial lives. The piggy-bank icon is the easy part.