Checklist · Confidentiality & client information
Protocol before using a self-learning generative AI tool (United States)
A check before anyone puts information relating to a representation into a generative AI tool that may learn from inputs or reveal them in later output, including to someone behind an ethical wall.
- Kind
- Checklist. Steps or checks to work through and tick off.
- Where it applies
- United States
- AI use
- Yellow You and the tool draft together, with you directing. You check every statement you rely on.
- Law stated as at
- 22 October 2026
- Status
- Maintained. Checked on 22 October 2026 against the 5 authorities it cites, for United States.
- Version
- 1
- Source
- Written for this library
When to use it
Use at the point of input, including a paste into a chatbot and a connector that can see a file. Use the tool-specific client consent if the check shows that information relating to a representation will be input. Do not use this protocol as the consent.
Used for these tasks
- Decide whether client material can go into a tool
- Build an inventory of tools that can see client material
- Compare a provider's terms on storage, training and subprocessors
- Build an upload checklist for a matter before any file goes into a tool
Each task page says how far AI may help with the task and what the lawyer must review.
How AI may be used
A tool may read this checklist against the firm's register and say which row is incomplete. It may not decide that input is allowed, and it may not do the input.
AI may
- Compare the tool name a person proposes to use with the name on the firm register and report a mismatch
- List checklist lines that are blank
- Point to the consent date in the file if one is already recorded for this tool and this client
AI must not
- Must not decide that information is not information relating to a representation
- Must not treat "not stated" in the provider's terms as a finding that the tool does not train on inputs
- Must not input the information, summarize the document inside the tool, or send the document to the provider in order to complete the check
- Must not waive an ethical wall because both matters are inside the firm
The lawyer checks
- The person completing the check is a lawyer, or a staff member the supervising lawyer has assigned, and a lawyer signs it before Class B input
- Self-learning status comes from the approval record, and "not stated" is treated as self-learning
- If an ethical wall exists, the check stops unless the wall still holds after the input and any later output
- Client consent, where required, is the separate signed consent for this tool, not this checklist
Method: how the Green, Yellow and Red levels are decided. This is guidance for planning, not legal advice.
Read the full text
The full text is available after you log in. It is free. When you log in you can:
- read the whole precedent, with its drafting notes beside it;
- copy it, or download it as a Word file or Markdown;
- add it to your firm’s knowledge base, so colleagues can find it;
- let the AI tools you connect read it from that knowledge base.
The summary, the law it rests on and the AI-use guidance above stay public. Privacy.
More in Confidentiality & client information
- AI confidentiality rules for staff (short form of the firm AI policy)
- Information barrier protocol for one matter
- Waiver of a concurrent conflict of interest (United States)
- Checklist before any file on a matter goes into a tool
- Staff and contractor AI confidentiality undertaking and acknowledgement
- AI vendor confidentiality and privilege questions (short form of the vendor due-diligence questionnaire)
- Inventory of systems that can see client material
- Staff undertaking on unapproved generative AI tools (United States)
- Engagement clause: client authority for AI processing of confidential information (rider to the AI engagement clauses)
- Comparison of a provider's terms on storage, training and subprocessors
- Check before using a hypothetical in a generative AI tool (United States)
- AI data incident steps (confidentiality short form of the AI incident response procedure)
- Confidentiality undertaking for an external reviewer or expert
All precedents for Confidentiality & client information · Tasks and AI use in this area