Is AI-Generated Legal Work Ethical? ABA Rules in 2026

Can Lawyers Ethically Use AI? Yes, With Guardrails

Lawyers can ethically use AI tools like Claude. The American Bar Association settled the threshold question in Formal Opinion 512 (July 29, 2024): generative AI is permissible in legal practice when attorneys comply with their existing professional responsibility obligations. A growing number of state bars have issued their own opinions or guidance, and the trend is permissive but cautious. The question in 2026 is not whether lawyers can use AI but how, and bar associations increasingly treat understanding AI as part of the competence obligation itself.

Formal Opinion 512 requires attorneys who use AI to:

• Maintain competence: have a reasonable understanding of the capabilities and limitations of the AI tools you use
• Protect client confidentiality: use AI platforms with proper data protections
• Supervise AI use and verify output: apply an appropriate degree of independent verification or review to AI output, and, if you manage or supervise others, set clear firm policies on AI use and make sure lawyers and staff are trained on it
• Communicate with clients: tell clients about AI use when the circumstances call for it, for example when the client asks, before you input information about their matter into a self-learning AI tool, when AI use is relevant to the basis or reasonableness of the fee, or when AI output will influence a significant decision in the representation

The rest of this guide maps those duties to the specific Model Rules and the state bar guidance built on top of them.

ABA Model Rule 1.1: Competence Now Includes AI Literacy

The duty of competence under ABA Model Rule 1.1 has been interpreted to include an obligation to understand the technology tools used in practice. Comment 8 to Rule 1.1, adopted in 2012, states that lawyers should "keep abreast of changes in the law and its practice, including the benefits and risks associated with relevant technology." In 2026, this means attorneys who use AI need a reasonable understanding of the capabilities and limitations of the AI tools they use, particularly the hallucination problem; Formal Opinion 512 says lawyers need not become AI experts.

Formal Opinion 512 also notes that emerging technologies may eventually become ubiquitous in legal practice and establish conventional expectations regarding the duty of competence, as other new technologies have done over time.

ABA Model Rule 1.6: Confidentiality in the AI Context

Rule 1.6(c) requires attorneys to make "reasonable efforts to prevent the inadvertent or unauthorized disclosure" of client information. When using AI tools, this means:

• Plan selection matters: Free and consumer-grade AI plans may use your inputs for model training. Claude's Team and Enterprise plans do not train on your data by default, and Enterprise adds custom data retention controls.
• Assess the risk before inputting client data: Formal Opinion 512 calls for a fact-specific evaluation of the risk that information entered into an AI tool will be disclosed or accessed, and the client's informed consent before you input information relating to the representation into a self-learning tool.
• Firm policies required: Formal Opinion 512 says managerial lawyers must establish clear policies on the firm's permissible use of AI; a written policy naming approved tools and data-handling rules is a practical way to meet that duty.
• Vendor due diligence: Just as you would vet any third-party service provider, review the AI provider's data handling practices.

ABA Model Rule 5.1/5.3: Supervision of AI Output

Rules 5.1 and 5.3 require attorneys to supervise the work of subordinates and nonlawyer assistants. Formal Opinion 512 applies these duties to generative AI: managerial lawyers must establish clear policies on the firm's permissible use of AI, supervisory lawyers must make reasonable efforts to ensure that lawyers and nonlawyers comply with their professional obligations when using AI tools, and the vendor due diligence used for outsourcing applies to AI providers. In practice, the responsible attorney must review, verify, and approve AI-generated work product before it is used in any representation. This means:

• Every AI-generated document must be reviewed before use
• Every citation must be independently verified
• Every legal conclusion must be assessed by a licensed attorney
• The attorney, not the AI, makes all strategic decisions

The Hallucination Problem: The Risk Behind the Supervision Duty

AI models, including Claude, can generate plausible but fabricated case citations, and courts have sanctioned attorneys for filing briefs with fake citations (see, for example, Mata v. Avianca, Inc. (S.D.N.Y. June 22, 2023)). Many sanctions cases share the same fact pattern: a lawyer submitted AI-generated citations without checking them. The fix is simple and non-negotiable. Verify every citation in Westlaw, Lexis, or the reporter itself, and treat AI output as a draft: useful, but reviewed before it reaches a client or a court.

ABA Model Rule 1.4: Client Communication About AI

Rule 1.4(a)(2) requires attorneys to reasonably consult with the client about the means by which the client's objectives are to be accomplished, and Rule 1.4(b) requires explaining matters to the extent reasonably necessary for the client to make informed decisions about the representation. Under Formal Opinion 512, lawyers must disclose AI use if the client asks or the engagement terms require it, must obtain the client's informed consent before inputting information relating to the representation into a self-learning AI tool, and must consult the client when AI use is relevant to the basis or reasonableness of the fee or when AI output will influence a significant decision in the representation. Formal Opinion 512 calls the engagement agreement a logical place to make AI disclosures, but general, boilerplate engagement-letter provisions are not enough to obtain informed consent for a self-learning tool.

Billing for AI-Assisted Work

Formal Opinion 512 is explicit that lawyers who bill hourly must bill for their actual time, not for the hours the task would have taken manually. If Claude compresses a four-hour review into forty minutes of attorney time, billing four hours misrepresents the work performed. Flat and contingent fees do not avoid the issue: the opinion applies the Rule 1.5(a) reasonableness factors to them too, and notes that if AI lets a lawyer finish much more quickly, charging the same flat fee as without the tool may be unreasonable. Lawyers also may not bill clients for time spent learning a generative AI tool they will regularly use for clients.

State Bar Guidance: Key Trends Across Jurisdictions

The state bar opinions and guidance issued through 2026 share common themes: review and verify AI output before it reaches any court, protect client data to the same standard as any third-party service, and remember that attorneys bear personal responsibility for all work product, with or without AI. Highlights by jurisdiction:

• California: The State Bar's Practical Guidance for the Use of Generative AI in the Practice of Law was originally published in November 2023, and revisions approved in May 2026 reflect developments including agentic AI. It covers competence, confidentiality, supervision, billing, and candor.
• New York: The New York State Bar Association's Task Force on Artificial Intelligence report was approved in April 2024, and the New York City Bar issued Formal Opinion 2024-5 (August 2024), which favors guardrails over new AI-specific rules.
• Florida: Ethics Opinion 24-1 (January 19, 2024) permits generative AI use if the lawyer protects client confidentiality, reviews AI work product as they would a nonlawyer assistant's, avoids inflated billing, and follows advertising rules for AI chatbots.
• Texas: The Professional Ethics Committee's Opinion 705 (February 2025) emphasizes that attorneys remain personally responsible for all work product regardless of AI involvement.
• New Jersey: The Supreme Court's Preliminary Guidelines (January 2024) apply existing rules to AI use and do not impose a blanket duty to disclose AI use to courts or clients.

Court Rules and Standing Orders on AI

A growing number of individual judges have issued standing orders, and some courts have adopted local rules, addressing AI use in court filings. Ropes & Gray reported in May 2026 that its AI Court Order Tracker catalogs over 550 standing orders, local rules, and decisions across state and federal courts. Requirements vary by court and by judge, and Formal Opinion 512 notes that some courts have responded by requiring lawyers to disclose their use of generative AI. The opinion advises lawyers to consult the applicable court's local rules, and in practice that means checking each judge's standing orders as well.

Building an Ethical AI Framework for Your Firm

Formal Opinion 512 requires managerial lawyers to set clear AI policies; we recommend putting yours in writing and covering approved tools and plans, data handling procedures, output verification protocols, client disclosure standards, court disclosure compliance, and billing guidelines. For a practical starting point, see our complete guide to using Claude for legal work, which includes an AI ethics framework. To understand how AI is being adopted across the profession, read How Law Firms Are Using AI in 2026.