Anthropic lets Claude read your bank account with new Personal Finance tool — but it cannot move a cent
Anthropic has launched Personal Finance in Claude, a read-only beta that links US bank and card accounts through Plaid so users can ask where their money goes — while the chatbot stays unable to move, manage or invest a cent.
Anthropic has given its chatbot Claude a window into its users' bank accounts. The company announced on Thursday that a new tool called Personal Finance will let people connect their bank and credit card accounts and ask Claude plain-language questions about where their money goes, in the clearest sign yet that the race between the big AI chatbots is moving from answering questions to managing everyday life.
The feature is rolling out in beta this week to paying subscribers in the United States on Claude's Pro, Max and Ultra plans, Anthropic said in a blog post. It works on Claude on the web, on desktop and on iPhone and Android, and it is limited for now to US banks. Free, Team and Enterprise subscribers are not included, and there is no date yet for users outside the United States.
The connection runs through Plaid, the service that already links bank accounts to thousands of finance apps. Users sign in to their bank through Plaid, and Claude never asks for a bank password and never sees it. Once linked, Claude can answer questions such as where the month's money went, which bills are coming up, or whether a planned trip is affordable, drawing on balances, credit limits and up to two years of transaction history.
The most important limit is what Claude cannot do. Anthropic says access is strictly read-only: the chatbot cannot transfer money, pay a bill, cancel a subscription or place a trade, and the feature is not designed to give investment advice. The company also says Claude sees only a narrow slice of each account — the bank's name, the account name and type, the last four digits of the account number, balances, credit limits and transaction history — and nothing else.
Anthropic is also drawing a line on advertising. The company says Personal Finance carries no ads and no sponsored recommendations, and that it does not sell its own credit cards, loans or savings accounts. That matches the position Anthropic set out in February, when it said Claude's answers would not be influenced by advertisers. The contrast matters in a market where free finance tools are often paid for by steering users towards products.
In its own demonstration, Anthropic showed a user asking where the month's money had gone. Claude totalled $4,309.37 across the connected accounts, with more than half going to rent and utilities and $1,024.51 to food and drink. In a follow-up, it listed four upcoming payments and $1,393.22 left before payday. The figures come from the company's own example rather than an independent test, but they show the kind of everyday arithmetic the tool is built for: totals, categories, due dates and what is left over.
The launch puts Claude roughly six months behind its biggest rival. OpenAI introduced similar personal finance tools in ChatGPT in May, also built on Plaid, and Perplexity shipped its own finance integration in April. All three now do broadly the same job — read a bank feed and answer questions about it — which means the contest has shifted to trust: whose limits users believe, and whose answers they can check.
Anthropic has been unusually direct about the weak point. It warns that Claude can make mistakes in its answers, and tells users to ask which transactions sit behind any number and to check important figures with their bank. Its help pages add that balances can be a day or two old, and that an account linked through two different logins can be counted twice if the bank presents it differently. For a tool that touches people's money, those are the details that decide whether it is useful or misleading.
The move is part of a wider push by Anthropic into specialised versions of Claude. Earlier this year it launched Claude for Legal and Claude for Financial Services, aimed at professional users, and Personal Finance is its first step into the personal money of ordinary subscribers. The company has also invited banks, wealth managers and financial services firms to help it work alongside licensed professionals on larger decisions, suggesting it sees the feature as a starting point rather than a finished product.
The timing is also significant. Reuters reported that the rollout comes ahead of Anthropic's highly anticipated initial public offering, as the company looks to show that Claude can grow beyond a general-purpose chatbot into services people use every day. A finance tool that people open weekly is worth far more to that story than one they try once and forget.
What happens next will depend on how far Anthropic takes it. For now the beta is gradual, limited to paying US subscribers, and capped at 16 bank links. If the rollout goes smoothly, the obvious questions are when it reaches other countries, whether it gains the ability to act on what it finds — paying a bill, for example, with the user's permission — and whether rivals answer with their own read-only promises. Until then, Claude can look at your money, add it up and talk about it. It just cannot touch it.
More in Technology & AI
Photo: Meta Platforms headquarters in Menlo Park, California (file image)
Technology & AIMeta opens video chats with its AI agent Muse to everyone in the US as Muse Voice goes live
Meta has switched on Muse Voice for all users in the United States, letting them call and video chat with its personal AI agent through a real-time voice conversation and an expressive animated avatar.
Photo: Tux, the Linux penguin mascot (file image)
Technology & AIOpenAI's AI Agents Build 948 Working Root Exploits for the Linux Kernel in a Single $1.2 Million Run
OpenAI says an autonomous pipeline built on an upcoming cyber model reproduced 1,724 Linux kernel vulnerabilities and turned 948 of them into working root exploits, each with a candidate patch, for $1.21 million in API credits.
Photo: The TikTok app on a smartphone screen (file image)
Technology & AIPTA dismisses fake TikTok ban notification as ‘false and misleading’, says no order issued
The Pakistan Telecommunication Authority said on Thursday that no notification banning TikTok had been issued, after a fabricated document claiming a nationwide suspension spread across social media.