Anthropic recently rolled out a new feature that lets its Claude AI analyze spending directly from linked U.S. bank accounts—a move that immediately drew sharp criticism from investor Anthony Pompliano.
Anthropic rolled out a personal finance update for Claude on Thursday, enabling users to sync their bank accounts so the chatbot can show recent spending, upcoming bills, and remaining balances.
Along with the launch came a serious claim from crypto investor Anthony Pompliano, who alleged that Anthropic attempted to extract insider details from his competing startup, Silvia, months before the feature went live.
Currently rolling out in beta for U.S. subscribers across Claude’s Pro, Max, and Ultra tiers, the new Personal Finance tool relies on Plaid to connect bank and credit card accounts securely, giving the AI access to financial data without exposing login credentials.
Once set up, Claude breaks down spending, catches duplicate charges or unexpected fees, clarifies interest costs, and generates visual charts or payment schedules. In one demo provided by Anthropic, a subscriber asks where their money went this month, then quickly checks how much cash remains before the next paycheck arrives.
“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.
Once connected, the AI can view current balances, credit limits, account types, the last four digits of account numbers, and up to two years of transactions from approved accounts. According to documentation released by Anthropic, conversations within the Personal Finance hub are excluded from model training, though users are advised to double-check AI-generated figures against actual bank records as balances can occasionally lag.
Questions also remain about how Anthropic handles user financial data beyond model training, an ambiguity that carries huge stakes given how immensely valuable this information is to big data companies.
Pompliano accuses Anthropic of fishing for secrets
Shortly after Claude’s announcement, Pompliano—whose startup Silvia frames itself as a personal AI chief financial officer—voiced his frustration in a detailed post on X. He claimed that Anthropic team members started reaching out to Silvia engineers back in April, ostensibly to discuss the company’s trajectory and computing requirements.
“Anthropic was hunting for proprietary details about our architecture and user traction,” he alleged. Pompliano further detailed a steak dinner with Anthropic representatives, where the Silvia team was informed that the bill was being covered directly by CEO Dario Amodei.
Although his engineers refused to share sensitive details, Pompliano noted that an insider at Anthropic tipped off his team about the impending release just days before it went public.
“Today, Anthropic debuted a Claude wrapper that poses as a personal finance tool,” Pompliano stated. He added that Silvia slashes computing expenses by nearly 97% by running custom models and hardware rather than relying on frontier AI systems, suggesting the startup depends on a local model tailored specifically to its needs.
Even so, Anthropic is hardly the first general-purpose AI developer to venture into personal finance. OpenAI rolled out similar bank integrations for ChatGPT back in May, relying on Plaid for read-only account access, while agents like Hermes, Open Claw, or Grok Bot can perform comparable tasks when configured with the right skills, permissions, and instructions.
Anthropic frames Personal Finance as just an initial step, noting that potential partnerships with licensed advisors and regulated financial institutions are actively being explored. For the moment, access to the beta remains limited to U.S. subscribers on Claude’s three paid tiers.
