Anthropic launches Claude Personal Finance, chasing ChatGPT for your bank data

Anthropic introduced Claude Personal Finance on November 5, 2026. The US beta allows paying users to link their bank accounts and cards via Plaid and ask Claude questions about their expenditures. ChatGPT already provides similar functionalities, and thus puts both assistants in a race to capture the consumer’s attention regarding any query related to finances.

From banks’ perspective, the rivalry bears more implications. Using a chatbot to get information on any fees or debts could influence a customer’s decision about whether to access their banking account apps or not.

What Claude can and can’t see

Before using Personal Finance, users need to open it in Claude’s sidebar. They need to click on the “Get started” option to follow directions to link accounts via Plaid. Now users can login to banking services using Plaid and choose which accounts to share. After they are connected to their bank, Claude will see bank and account names, types of accounts, last four digits of the account numbers, balances, credit limits, as well as up to two years of transaction history. This information is used to analyze expenditures, identify duplicate payments and rising invoices, calculate paid interest, and generate charts.

Claude is only able to view the above information. It cannot initiate transfers nor handle money transactions of any type. Anthropic clarifies that Claude was not meant for investment advice. Personal Finance does not have any sponsored advertisements or recommendations. Data from conversations are not used for training models at Anthropic. Users can easily disconnect their accounts whenever they wish.

For now, the US beta is available to Claude Pro, Max and Ultra subscribers. The plans cost $20, $100, and $200 per month, respectively, when billed monthly through the web.

Playing catch-up to ChatGPT

Claude Personal Finance is currently in beta for paying US subscribers. It arrives nearly six months after OpenAI introduced its personal finance feature in ChatGPT on May 15, 2026. OpenAI initially offered the feature as a preview for US Pro subscribers. Access expanded to Plus on June 25, then began reaching Free and Go users on October 2. ChatGPT has since added weekly insights, investment dashboards, credit monitoring, and voice support.

OpenAI reports that ChatGPT receives over 200 million inquiries about money every month. Although this number speaks volumes about how widely it is used for finance, it does not provide any indication of how many users have connected their accounts.

Both Claude’s Personal Finance and ChatGPT’s personal finance feature rely on Plaid, which connects more than 12,000 financial institutions. It provides the account connections, transaction data, and user permissions behind these tools. Plaid stands to benefit regardless of which chatbot consumers prefer.

The real prize is the customer relationship

For banks, the bigger concern is losing their direct connection with customers. McKinsey warned on April 30, 2026, that AI agents could become consumers’ main banking interface. Its research found that 23% of surveyed consumers used generative AI for financial tasks at least once a month.

A May 18 Bain analysis offers a glimpse of how this could affect banks. Researchers examined 6,000 consumer prompts in Australia and found that roughly one-third of banking queries concerned home loans. AI-generated answers were helping customers narrow their choices before visiting bank websites.

Interest in AI-powered financial assistance extends beyond Australia. A 2026 peer-reviewed survey of 2,170 South Korean adults found that 67.8% had used large language models for at least one financial task. However, the finding reflects one national sample, not global adoption.

Trust, accuracy and regulation could slow it

Growing use does not necessarily mean consumers are ready to trust AI with their finances. Deloitte found that 79% of surveyed US bank customers trusted bank websites for product research. Only 49% said the same about generative AI. Another 83% were concerned about AI acting on their finances without permission.

Accuracy raises further questions. The OECD warned in 2026 about incorrect answers, bias, privacy risks, and hidden commercial influence. It also cautioned that AI could blur the distinction between financial education and personalized advice.

Anthropic itself advises users to verify important figures with their banks. Beyond accuracy and privacy, expansion will depend on local rules governing financial data and advice. Those requirements differ across countries and could slow Claude’s rollout beyond the US.