
Anthropic introduces a beta personal finance tool for U.S. subscribers while facing allegations of soliciting trade secrets from startup Silvia.
AI-generated summary
Anthropic is expanding Claude's capabilities into personal finance, following a similar move by OpenAI with ChatGPT earlier this year.
Anthropic launched a personal finance feature for Claude on Thursday, letting users connect bank accounts and ask the chatbot what they have spent, what bills are due, and how much money they have left.
The announcement came with an accusation from crypto investor Anthony Pompliano, who says Anthropic tried to get inside information from his competing startup, Silvia, months before the rollout.
The new feature, called Personal Finance, is rolling out in beta to U.S. subscribers on Claude's Pro, Max, and Ultra plans. Users can link bank and credit card accounts through Plaid, which handles the bank login and gives Claude access to financial data without sharing account passwords.
Claude can then summarize spending, spot duplicate charges or unexpected fees, explain interest payments, and build charts or bill calendars. Anthropic's example shows a user asking where their money went this month, then following up to find out what remains before payday.
"Access is read-only, and Claude can't move, manage, or invest your money," Anthropic wrote on X. "Like the rest of Claude, Personal Finance has no ads or sponsored recommendations," the company added.
The assistant can see balances, credit limits, account names and types, the last four digits of account numbers, and up to 24 months of transactions from accounts users approve. According to Anthropic's documentation, conversations inside the Personal Finance hub are not used to train its models, but account balances may lag and AI-generated figures should be checked against bank records.
Also, how Anthropic tracks its users’ financial data and uses it beyond model training is not completely clear—and this is extremely valuable information for big data companies.
Pompliano accuses Anthropic of fishing for secrets
Pompliano, whose Silvia product bills itself as a personal AI chief financial officer, posted a lengthy complaint on X shortly after Claude's announcement. He alleged that Anthropic personnel began approaching Silvia engineers in April, supposedly because they were interested in the startup's growth and computing needs.
"Anthropic was fishing for details on how we built our product and why it was so popular," he wrote. Pompliano also described an alleged steak dinner involving Anthropic representatives, claiming someone told the Silvia team that the meal was on CEO Dario Amodei's tab.
His engineers declined to divulge sensitive information, he claimed, adding that someone at Anthropic warned his team about the upcoming launch days earlier.
"Today, Anthropic launched a Claude wrapper that is masquerading as a personal finance product," Pompliano wrote. He also claimed Silvia's computing costs are roughly 97% lower when using its own models and hardware instead of frontier AI models, which most likely mean they are using a local model, likely finetuned for the startup’s specific purpose.
That said, Anthropic was not the first general-purpose chatbot maker to move into personal finance, either. OpenAI launched similar bank connections for ChatGPT in May, also using Plaid for read-only access to financial information, and Agents like Hermes, Open Claw, or Grok Bot can also do pretty similar tasks given the proper skills, permissions, and instructions.
Anthropic says Personal Finance is a first step and is exploring partnerships with licensed professionals and regulated financial firms. For now, the beta is restricted to U.S. customers on three paid Claude plans.

OpenAI promised that neighbors of its future AI campuses will not pay for their electricity needs, pledging public energy and water figures and limits on secrecy agreements with officials, two days after voters backed data center restrictions in multiple states.

The U.S. Department of Justice has announced the Super Intelligence Task Force (DOJ-SITF), led by Associate Attorney General Stanley Woodward and Associate Deputy Attorney General Stephen Weiss, to police misuse of advanced AI by foreign governments, criminal organizations, companies, and individuals. Attorney General Todd Blanche said the task force will work with SI Force Chair Jay Clayton to prosecute illegal SI use and push AI developers to strengthen internal controls.

OpenAI reports its autonomous security agents identified 948 distinct Linux kernel vulnerabilities, developed 1,169 working exploits, and drafted fixes for approximately $1.2 million in API credits, raising concerns about AI-assisted cyberattacks.

OpenAI CEO Sam Altman defended his comments on DRAM allocation, stating he supports increased memory production after facing backlash for suggesting that personal computer memory usage is inefficient compared to cloud-based AI data center needs.

Google announced that AI products will be disabled by default for K-12 student accounts unless administrators enable them, but schools that want to ban AI entirely must also disable Google Search. The changes take effect in early 2027 amid antitrust scrutiny and criticism from politicians including New York City Mayor Zohran Mamdani and former FTC chair Lina Khan.

OpenAI announced its Grid Protectors Program, allowing future AI facilities to reduce electricity consumption during grid emergencies, with a Georgia facility covering 3,200 megawatts and up to 1,000 megawatts available for temporary reductions. The company is also rebranding data centers as "AI compute campuses" amid community opposition.