Can Anthropic see my clients' data?
By Dr. Leigh Coney, Founder of WorkWise Solutions · Updated October 20, 2026
Short answer: Anthropic's systems process everything you send to Claude, so your clients' data does pass through Anthropic. Its people are a separate question. On Team and Enterprise, Anthropic's Data Processing Addendum says it processes customer personal data only to provide the service and on your documented instructions, gives staff least-privilege access, logs employee access and binds staff to confidentiality. Anthropic's documents also describe when content is kept longer or reviewed: chats its safety systems flag, conversations someone sends as feedback, safety work on its covered models, and disclosures the law requires. Ask for its staff-access rules in writing for your vendor file.
Two questions hide in this one. Can Anthropic's systems see the data? Yes: Claude has to read a household's statements to summarize them. Can Anthropic's people read it? That depends on your plan and your contract.
For consumer plans, Anthropic answers in one sentence. Its privacy center says, of Free, Pro and Max: "By default, Anthropic employees cannot access your conversations unless" the user shares them as feedback, or a review is needed to enforce its Usage Policy. Then only "designated members of our Trust & Safety team may access this data on a need-to-know basis".
For Team and Enterprise, Anthropic publishes no equivalent sentence. The protections sit in the contract. Its Data Processing Addendum says Anthropic will only process customer personal data "to provide or maintain the Services, and in compliance with Customer's documented instructions". It adds that everyone who processes that data has a duty of confidentiality, that staff get least-privilege access with regular reviews, and that "Anthropic logs employee access and actions within Anthropic's systems".
Some content stays longer than your settings would keep it:
- Flagged chats. If automated safety systems flag a chat as a Usage Policy violation, Anthropic keeps the inputs and outputs for up to 2 years.
- Feedback. A thumbs-up or thumbs-down sends Anthropic the whole conversation, kept for up to 5 years. Owners can switch ratings off.
- Covered models. For the models Anthropic calls covered (Fable 5 and 5.1, Mythos 5 and 5.1), it keeps prompts and outputs for 30 days for safety work. Its policy adds: "By default, no Anthropic personnel can read your retained conversations." Human review happens only through a controlled path, such as content its automated systems flag, by a small set of approved reviewers.
- The law. Anthropic's commercial terms let it disclose confidential information when a law or court order requires it, with prompt notice to you unless it is prohibited from giving it.
Your own side counts too. On Team and Enterprise, the Primary Owner can export the organization's data, which can include conversations and uploaded files. Decide who holds that role, and write it down.
Two routes shrink the question. Schwab masks account numbers, Social Security numbers and dates of birth before data reaches Claude. And Anthropic's docs say that when Cowork runs on your own Amazon Bedrock or Google Cloud account, it sends no prompts or completions to Anthropic. If you pick a covered model there, 30-day retention has to be switched on, and Anthropic says that data stays in your AWS or Google Cloud environment.
For your vendor file, ask Anthropic to state its staff-access rules for commercial plans in writing. The DPA terms are a strong start, and a direct answer is better evidence than an inference.
I set it up for RIAs. Through WorkWise Solutions, the Claude for Financial Advisors setup covers the plan decision, the connectors, your firm's rules in the skills and archive capture, at a fixed price: $10,000 for firms of 1 to 10 people or $20,000 for 11 to 25, in two to four weeks.