Guides

Written for people who have to make a decision this afternoon: whether a document can go into a chatbot, what “de-identified” actually means, and how to check a privacy claim instead of believing it.

• How to verify a “private” tool really uploads nothing

  Four checks anyone can run in a browser to test whether a redaction or conversion tool actually processes your data locally — including on this site.

• What ChatGPT, Claude and Gemini actually do with what you paste

  Training, retention, human review and legal holds — what the providers state as of July 2026, where the consumer and business answers differ, and how to check.

• The placeholder round trip: keeping an AI answer usable after redaction

  How to scrub a prompt without wrecking the answer — token choice, prompt phrasing, what models do to placeholders, and how to restore the real values afterwards.

• Redaction vs anonymisation vs pseudonymisation

  Three words used interchangeably in marketing and precisely in law. What each one means, which one this tool does, and when only real anonymisation will do.

• Is it safe to paste client data into ChatGPT?

  The honest answer, the four questions that actually decide it, and what to do when the answer is no — written for people who have to decide today.

• The 18 HIPAA identifiers, in plain English

  HIPAA Safe Harbor lists 18 identifier types to remove — plus a condition most summaries omit. What each one covers, and why removing all 18 still is not always enough.