All answers
Security & compliance Open question — the Lab is working on this

Does using AI tools like Claude with client tax return data raise IRC 7216 concerns, even if the data is scrubbed or masked?

18:09From the June 2 call · Beginner Lab: Claude Basics, Custom GPTs, and Practice Wins

One member referenced a comment from Brandon Hall on a social media post, noting there is fine print suggesting that even if you scrub or mask sensitive client information before feeding it to an AI tool, you may still be 'doing something' that implicates IRC 7216 -- Brandon's recommendation was to have an attorney help develop a written policy before proceeding. The member described testing a workflow with Claude: feeding it a prior-year tax return without exposing sensitive details, having it generate a checklist of documents needed for the current year, turning incoming client documents into organized work papers, doing a first-pass review, and producing a plain-language summary of the return for a non-tax person. When shared publicly, some people agreed with the approach and others pushed back. The discussion ended on an open question about how to actually verify a tool's security claims (e.g., if Claude says it 'can't see anything,' how do you confirm that's true), which was not resolved in this segment.

The full answer is members-only

Membership is free. Apply and get this answer, the recording, and the rest of the library.

Apply to join

Already a member? Sign in