Is it an ethics violation to use ChatGPT if the client data is redacted?

Where redaction fits in the guidance

Law Society of Ontario: "redact any identifying, sensitive, or confidential information", and where anonymizing cannot protect confidentiality or privilege, explain the risks and obtain informed consent. Law Society of British Columbia: "Ideally, client confidential information, including any information identifying the client, would be omitted." Law Society of Alberta: "Never include confidential or potentially identifying information in prompts."

Florida Bar Ethics Opinion 24-1: "It is recommended that a lawyer obtain the affected client's informed consent prior to utilizing a third-party generative AI program if the utilization would involve the disclosure of any confidential information." It adds that if the use "does not involve the disclosure of confidential information to a third-party", consent under Rule 4-1.6 is not required. Removing names does not necessarily make the remaining facts non-confidential, which is why redaction narrows this question rather than answering it.

What redaction does not fix

• Residual facts: a unique deal, a small town or a rare job can identify the client without a name
• Missed details: automatic detection misses some identifiers, especially in scans
• Accuracy: you remain responsible for what the AI produces and what you file
• Other duties: supervision, fees and candour to the court are untouched by redaction

A defensible practice

Remove direct identifiers, reduce quasi-identifiers the task does not need, read the copy the AI will see, keep the key off the AI platform, and follow your law society's consent guidance where confidentiality still cannot be protected. Record what you did.