Frequently Asked Questions

Can lawyers use ChatGPT for legal work?

Yes — but only if identifying client information never reaches OpenAI's servers. ABA Model Rule 1.6 requires reasonable efforts to prevent inadvertent disclosure of client information. Submitting a prompt containing a client's name or case details is a disclosure under that rule. Attorneys can use ChatGPT safely by manually redacting before submission, or by using an on-device tool like Naosuu that handles redaction automatically. Read more →

Is it ethical for attorneys to use ChatGPT?

It depends on how it's used. ABA Formal Opinion 512 (2023) permits AI use for legal work but requires attorneys to understand the tool's data practices and protect client confidentiality before any submission. Bar associations in more than thirty states have issued parallel guidance — none prohibit AI, all require a safeguard that prevents client data from reaching the AI server. Read more →

What does ABA Rule 1.6 say about using AI tools?

Rule 1.6(c) requires attorneys to make reasonable efforts to prevent the inadvertent or unauthorized disclosure of information relating to a client representation. While the rule predates AI, it covers any disclosure by any means. Submitting a prompt to ChatGPT that contains client information is a disclosure under this rule. Read more →

What information should lawyers never share with ChatGPT?

Any information that would identify your client, the matter, or the opposing party: client names, case numbers, docket references, contact and location information, financial figures tied to a specific matter, government-issued identifiers, and medical information. Combination risk is commonly overlooked — a city, injury type, and distinctive financial figure together can be identifying even when no name is present. Read more →

Does OpenAI store ChatGPT conversations?

Yes, by default. OpenAI logs conversations and may use them to improve its models. Opting out in settings reduces training use but does not eliminate server-side processing at the time of submission. That transit is the exposure — on-device redaction eliminates it by replacing identifying content before the prompt reaches OpenAI's servers.

How do I use ChatGPT for legal drafting without exposing client data?

The four-step approach: identify every element that would identify your client; replace each with a consistent placeholder like [CLIENT] or [CASE_NUMBER]; submit the redacted prompt; restore real values in the response. Naosuu automates steps 2–4 for every prompt you submit to ChatGPT, Claude, or Gemini. Read more →

Does this apply to Claude and Gemini as well?

Yes. The same Rule 1.6 analysis applies to any commercial AI service that receives a prompt containing client information — including Claude (claude.ai) and Google Gemini. Each platform processes prompts on its own servers. Naosuu protects all three automatically with the same on-device redaction workflow.

Can a firm ban ChatGPT to avoid the confidentiality issue?

They can, but the ban is routinely ignored. Attorneys use personal devices and personal accounts outside the firm's control. A technical safeguard that works wherever the attorney works is more effective than a policy that depends on consistent compliance under time pressure.