The Crossing Report — Issue #11

Your Engagement Letter Needs an AI Clause: The Compliance Checklist for Law Firms (2026)

Published April 20, 2026 · By The Crossing Report · 12 min read

This is educational content, not legal advice. Consult your professional liability carrier and state bar for jurisdiction-specific guidance.

Summary

ABA Formal Opinion 512 now sets the professional responsibility standard for AI use in legal practice. Multiple federal district courts require disclosure of AI in filings — and the list is growing. A small law firm using AI for drafting, research, or client communication without an updated engagement letter, an internal AI use policy, and court-filing disclosure language is carrying active liability exposure. This page covers the three documents every AI-using law firm needs to update, what those updates should include, and the action checklist for this week.

What Changed: ABA Opinion 512 and Court-Level Disclosure

The professional guidance phase is over. The compliance phase has started.

ABA Formal Opinion 512 has been in force since 2024, and it governs AI use for practitioners in ABA-member states. Opinion 512 is not a recommendation or a best-practices guide. It's the professional responsibility standard applied to three rules every attorney already knows: competence (Rule 1.1), supervision of non-attorney assistants (Rule 5.3), and client confidentiality (Rule 1.6).

Multiple federal district courts have issued standing orders requiring attorneys to disclose AI use in filings. The requirements vary by court: some require a certification that any AI-generated content was reviewed and verified by a licensed attorney. Others require disclosure of which tools were used. Check current federal and state court requirements at spellbook.legal/learn/state-bar-rules-ai-use, which maintains an updated state-by-state guide.

Key takeaway

A small law firm that is using AI and has not updated its engagement letters, created an internal AI use policy, and confirmed court-filing disclosure requirements for its jurisdiction is not behind on best practices. It is carrying active, documented professional responsibility exposure.

Three Documents Every Law Firm Using AI Must Update

Most law firms using AI are carrying one of three gaps. Fix these in order.

  1. 1.Engagement letter — Client-facing disclosure of AI use and oversight
  2. 2.Internal AI use policy — Who can use what, under what supervision
  3. 3.Court filing disclosure template — Ready-to-use certification language for filings in courts that require it

This is not a 6-month project. All three can be drafted in a single afternoon.

Engagement Letter AI Clause: What to Include

A reasonable engagement letter AI clause addresses four elements:

  • 1.Scope of use. Which categories of work may involve AI tools: legal research, document drafting, communication drafting, document review.
  • 2.Oversight. All AI-assisted work is reviewed and verified by a licensed attorney before use in client matters. The attorney of record retains professional responsibility for all work product.
  • 3.Confidentiality. The firm uses enterprise-grade AI tools that do not train on client data, consistent with client confidentiality obligations.
  • 4.Client opt-out option. Clients who prefer that AI tools not be used in their matter should say so.

Have your professional liability carrier review the language before finalizing. Many carriers now offer AI use policy endorsements or have guidance documents.

Internal AI Use Policy: The Three Rules

A firm-wide AI policy does not need to be a long document. The goal is to create the supervision structure that Opinion 512 requires.

  • •Rule 1 — Permitted tools. Define which AI tools attorneys may use without specific approval. Typically: enterprise accounts for Claude (Anthropic), ChatGPT (OpenAI), or purpose-built legal AI tools (Spellbook, Harvey, Lexis+ AI, Westlaw AI). Consumer-grade free tiers are not permitted for use with client data.
  • •Rule 2 — What requires attorney review before use. Any AI-generated text that will become work product must be reviewed by a licensed attorney before transmission or filing. The attorney who reviews it is professionally responsible for it.
  • •Rule 3 — What is prohibited. Submitting any AI-generated text as work product without attorney review. Using consumer AI tools with client data. Using AI for tasks requiring jurisdictional specificity without validating outputs against primary sources.

Action Checklist: This Week

Three concrete actions. Pick the first one you haven't completed.

  • 1.Update your engagement letter. Draft a 4-element AI clause using the structure above. Have a partner review it. Implement it for all new engagements immediately.
  • 2.Set your internal AI use policy. Write three rules: permitted tools, what requires attorney review, and what is prohibited. Circulate it to all attorneys and staff within the week.
  • 3.Check your courts. Go to spellbook.legal/learn/state-bar-rules-ai-use and check each court where you regularly file. Create a simple table: court name, disclosure requirement, and exact certification language required.

Premium Content

Engagement Letter Template + Court Disclosure Language

Get the fill-in engagement letter AI clause template, the internal AI policy one-pager, and court-specific disclosure certification language. Premium subscribers also get every future issue in full.

Free weekly digest. No spam. Unsubscribe anytime.

$19/month · Cancel anytime · First issue free

FAQ — AI Compliance for Law Firms

Q: Is ABA Opinion 512 legally binding?

A: ABA opinions are interpretive guidance, not statutory law. However, state bar disciplinary rules incorporate ABA model rules, and Opinion 512 interprets competence (Rule 1.1), supervision (Rule 5.3), and confidentiality (Rule 1.6) as applied to AI. A disciplinary complaint against an attorney for misusing AI will be evaluated against Opinion 512 standards in ABA-member jurisdictions. Treat it as binding for practical purposes.

Q: What happens if a court filing uses AI without disclosure?

A: Courts that require AI disclosure treat non-disclosure as a sanctionable offense. In several publicized cases, attorneys were sanctioned for submitting AI-generated content without disclosure or review — most famously in cases involving hallucinated case citations. A standing order violation can result in sanctions, dismissal of filings, or referral to the state bar.

Q: Do small law firms (under 10 attorneys) need an internal AI policy?

A: Yes, especially if attorneys are using AI tools individually without coordinated oversight. An internal policy doesn't need to be a 30-page document — three rules suffice: what tools are permitted, what requires supervisor review before use, and what is prohibited. The policy creates the supervision structure that ABA Opinion 512 requires.

Q: Which AI tools are safe to use with confidential client data?

A: Enterprise-grade tools with data processing agreements: Claude for Business (Anthropic), ChatGPT Enterprise (OpenAI), and purpose-built legal AI tools (Harvey, Spellbook, Lexis+ AI, Westlaw AI) that include BAA/DPA terms. Consumer versions of these same tools — free tiers, personal accounts — should never touch client data.

Q: How quickly do engagement letters need to be updated?

A: Before your next new client engagement. Existing engagement letters with ongoing matters are more nuanced — consider whether a letter amendment or client communication is needed for matters where you are actively using AI in ways the original engagement letter didn't contemplate. Your professional liability carrier may have guidance on this.

Sources & Further Reading

  • ABA Formal Opinion 512 — Professional responsibility standards for AI use in legal practice (2024)
  • Spellbook state-by-state AI disclosure guide — Updated federal and state court AI disclosure requirements
  • Clio Operate (Legalweek New York, March 2026) — AI-using firms reporting 40% shorter case lifecycles, 80% faster matter creation

Related Reading