Does Claude for Financial Advisors train on client data?
By Dr. Leigh Coney, Founder of WorkWise Solutions · Updated September 28, 2026
Short answer: Not on commercial plans. Anthropic's commercial terms for Team and Enterprise do not use your inputs to train its models, and Schwab says its agreement with Anthropic prohibits training on client data. Storage is a separate question: what Claude reads from your systems becomes part of the conversation and is kept under your organization's retention settings, so treat it like any other client record.
Here is how the data moves. The plugin's own files hold no client data. When a skill runs, Claude sends a query to the connected system and gets back only what it asked for. The records stay at the source, and the returned data enters the conversation.
Some partners add their own protection. Schwab masks account numbers, Social Security numbers and dates of birth, and lets the firm choose which users can use its connector.
On Enterprise, the Compliance API can export conversations and files to your archive, and admins control retention. The audit log records events such as sign-ins and file uploads for 180 days, without the content.
Personal accounts are the real risk. Anthropic's consumer plans let individuals choose whether their chats help train models, and a personal account sits outside your archive and your policies. Client work belongs in the firm's workspace, and your AI policy should say so in one sentence.
For Regulation S-P, treat Anthropic and each connector vendor as service providers with access to customer information, each with a vendor file.