AI Lawyer Ethics: Confidentiality, Verification, and Oversight

AI lawyer ethics means applying existing professional responsibility rules to AI-assisted legal work. The core duties—competence, confidentiality, supervision, and candor—do not disappear when an AI tool helps with research or drafting. They shift in form: a lawyer must understand what the tool does, keep client data out of the wrong places, and verify the output before relying on it. Tools like fenbs help legal teams record approved AI work rules and track who authorized each one.

The American Bar Association and state bar associations have issued guidance on AI, most anchored in existing Model Rules rather than new requirements. Understanding which rules apply, and how, is the first step before connecting any AI tool to legal work.

What professional responsibility rules cover

Several Model Rules are directly relevant. Rule 1.1 (Competence) requires understanding the technology used in practice, including its benefits and risks. Rule 1.6 (Confidentiality) governs what client information can be shared with outside services. Rule 5.3 (Supervision of Non-Lawyer Assistance) applies to AI tools that assist with legal work.

Some state bars have added AI-specific guidance. Check with your state bar for current opinions. The American Bar Association publishes formal opinions and ethics resources at https://www.americanbar.org/↗ (opens in a new tab). When in doubt, read what your jurisdiction has said before deploying a new AI tool on client matters.

The confidentiality risk in AI tools

Rule 1.6 requires reasonable measures to prevent unauthorized disclosure of client information. Before entering client facts into any AI tool, ask: where does this data go? Is it used to train the vendor's model? Can other users see it? Does sharing it require client consent?

Most major AI vendors offer enterprise or confidential-processing options that address some of these concerns, but the terms vary. Review the vendor's data-processing agreement, not just the marketing page. Write down which tools are approved for client data and which are not—that is a firm-level decision that belongs in your approved AI use rules.

Accuracy and the duty to verify

AI tools, including those designed for legal research, generate citations that may not exist. Model Rule 3.3 (Candor toward the Tribunal) prohibits presenting false authority. A lawyer who submits an AI-generated brief without verification is not protected by the fact that the AI created the error.

Every case citation, statutory reference, and regulatory quotation produced by AI must be verified against a primary source before use. Build this as a mandatory step, not an optional one. The guide to AI lawyer hallucinations covers verification techniques and common error patterns in legal AI research. Record the verification requirement as a firm rule so every lawyer and AI assistant in your practice knows it applies.

Supervision requirements for AI assistance

Rule 5.3 applies when AI tools assist with legal work in a way analogous to non-lawyer assistance. The supervising lawyer is responsible for the work product. Saying "the AI wrote this" is not a defense under supervision obligations.

Supervision means more than reviewing the final draft. It means understanding what the tool was asked to do, whether its method is suited to the task, and what could go wrong. For a legal research summary, that includes knowing the tool's known limitations with hallucinated citations and running an independent check.

Disclosure to clients

Whether to disclose AI use to clients depends on the jurisdiction and the nature of the use. Some ethics opinions require disclosure when AI substantially contributes to work product, particularly if client information is involved. Others frame disclosure as a matter of good practice.

Decide your firm's disclosure policy and record it clearly. A clear policy prevents inconsistent decisions across matters and gives clients a reliable answer when they ask. Document who made the policy decision and when, so it can be reviewed when guidance evolves.

How to record approved AI work rules

The most common ethics failure in legal AI use is not the tool itself—it is using the tool without agreed rules, so each lawyer makes individual judgment calls that may conflict. A firm-level record of approved tools, approved uses, data handling rules, and verification requirements gives every lawyer the same starting point.

When rules change—new bar guidance arrives, or a vendor changes its data terms—update the record and note what changed and why. An AI assistant connected to your practice management should read the current rules, not ones written six months ago that nobody updated.

How to do this with fenbs

Start with the decisions you have already made about AI use, even informally. Writing them down as firm rules is the most immediate improvement you can make.

1. Create a free account and set up a project for your firm's AI governance. Add a short description of the goal: approved rules for AI use in legal work.
2. Write the core decisions in Decisions and rules: approved tools, data handling limits, the verification requirement, and disclosure policy. Name the partner or ethics committee who decided each one.
3. Create a task for each rule that needs a scheduled review, such as checking new bar guidance or reviewing vendor terms. Assign a named owner.
4. When a rule changes, add a new decision that supersedes the old one and notes what changed. Keep the previous decision in the record so the reasoning is still available.
5. Connect your AI assistant over MCP from Integrations so it reads your approved rules at the start of every session before it assists with legal work.

fenbs does not connect to your case management system, read client files, or provide legal advice. It holds the governance record: the approved rules, the decisions behind them, and the tasks that keep them current.