As AI tools have become a practical part of legal practice, bar associations and courts have accelerated their guidance on what’s permissible, what’s required, and what crosses ethical lines. Understanding these rules isn’t optional — competence, supervision, and confidentiality obligations follow attorneys into their AI workflows.
This guide covers the most important ABA guidance, state bar developments, and court-specific rules attorneys need to know before integrating AI into their practice.
The ABA’s Position on AI in Legal Practice
The American Bar Association has addressed attorney AI use through several channels: formal ethics opinions, the Model Rules of Professional Conduct, and guidance from its standing committees.
ABA Formal Opinion 512 (2024) — Generative AI
The ABA’s most directly relevant guidance on AI is Formal Opinion 512, issued in 2024. The opinion addresses the use of generative AI tools (such as ChatGPT, legal AI products, and similar tools) in legal practice.
Key guidance from ABA Formal Opinion 512:
Competence (Rule 1.1): Attorneys must understand the capabilities and limitations of the AI tools they use. Using AI without understanding its outputs — including its potential to hallucinate citations, mischaracterize law, or generate incorrect analysis — can constitute a competence violation. The ABA notes that the duty of competence extends to technology used in legal practice.
Supervision (Rules 5.1, 5.3): AI tools used in legal work must be supervised as if they were a junior lawyer or non-lawyer assistant. Attorneys cannot abdicate professional judgment to an AI output. This applies equally to AI-generated research, AI-drafted documents, and AI-assisted contract review. Supervision means reviewing AI outputs for accuracy and professional judgment, not merely accepting them.
Confidentiality (Rule 1.6): Before submitting client information to any AI tool, attorneys must evaluate whether doing so creates risks to client confidentiality. Many commercial AI systems train on input data by default. Attorneys should use AI tools with clear data processing agreements that prohibit training on client data. Enterprise legal AI tools (Harvey, Clio Duo, Spellbook, Ironclad) typically have confidentiality commitments built into their terms of service. Consumer AI tools (ChatGPT’s free tier, Google Gemini’s free tier) may not.
Candor to Tribunal (Rule 3.3): If AI-generated content is submitted to a court, attorneys have an obligation to verify its accuracy. Submitting AI-hallucinated citations — citations to cases that don’t exist — has resulted in sanctions in multiple high-profile cases. Attorneys are responsible for the accuracy of what they file, regardless of how it was generated.
Fees (Rule 1.5): The ABA opinion notes that attorneys should consider how AI use affects fee arrangements. If AI dramatically reduces the time required for a task, billing clients for the time that would have been required without AI may create fee issues, particularly under hourly billing models. The opinion encourages transparency with clients about AI use in billing discussions.
State Bar Guidance: Key Developments
Several state bars have issued their own AI guidance, some more prescriptive than the ABA’s model rules approach:
California
The California State Bar has been among the most active on AI ethics. Its guidance (2024) addresses:
- Duty to supervise AI outputs
- Disclosure obligations when AI is used in client work (the bar recommends disclosure, though it stops short of requiring it in all cases)
- Prohibition on using AI in ways that violate existing confidentiality obligations
New York
The New York State Bar Association has issued guidance emphasizing:
- Competence requirements when using AI in legal work
- Supervision of AI-generated work product
- Confidentiality analysis before submitting client data to AI systems
- Caution around AI use in client-facing communications without review
Florida
The Florida Bar addressed AI ethics earlier than most state bars, noting in 2023 that existing rules of professional conduct apply to AI use without creating AI-specific exceptions. Florida has also addressed AI use in advertising — AI-generated marketing content must comply with attorney advertising rules.
Court-Specific AI Rules
Several federal courts and state courts have issued standing orders specifically addressing AI use in court filings:
Courts with certification requirements: A growing number of federal district courts require attorneys to certify whether AI was used in the preparation of court filings. These requirements vary — some require disclosure of any AI use, others require disclosure only if AI-generated language was used without human review.
Sanctions for AI hallucinations: Courts have imposed sanctions on attorneys who submitted briefs citing non-existent cases generated by AI. The most widely publicized examples involved ChatGPT-generated citations that no attorney verified before filing. These sanctions have increased attorney attention to AI verification obligations.
Best practice: Before filing in any federal court, check the court’s local rules and any standing orders from the assigned judge regarding AI use. The Judicial Conference of the United States has signaled that AI disclosure requirements will continue to expand.
Practical Compliance: What Attorneys Should Do
1. Choose AI tools with appropriate confidentiality protections.
Enterprise legal AI tools have contractual commitments not to train on your data. Consumer AI tools may not. Before submitting client information to any AI system, review the terms of service and, if necessary, execute a data processing agreement.
Recommended tools with enterprise confidentiality commitments:
- Clio (and Clio Duo) — enterprise security, SOC 2 Type 2 certified
- Spellbook — processes data in compliance with enterprise privacy standards
- Draftwise — enterprise security, iManage integration
- Harvey AI — enterprise data controls, does not train on client data
- Westlaw Precision / Lexis+ — established legal data providers with confidentiality standards
Avoid using free tiers of general AI tools (ChatGPT free, Gemini free) for client matters.
2. Review, don’t just use, AI outputs.
The ABA’s supervision requirement means AI output must be reviewed as you would review work from a junior associate — critically, for accuracy and professional judgment. Build review checkpoints into your AI workflow.
3. Verify all citations before filing.
If AI research output includes case citations, verify each citation independently before including it in a court filing. Legal research platforms like Westlaw and Lexis automatically ground AI responses in their verified databases, significantly reducing hallucination risk compared to general AI tools.
4. Consider disclosure to clients.
While ABA Opinion 512 does not require disclosure of AI use in all circumstances, disclosure is good practice — particularly for clients who have expressed concern about AI, in matters involving highly sensitive information, or when AI is being used in ways that affect billing.
5. Stay current on court-specific rules.
AI-related court rules are evolving rapidly. Check local rules and judicial standing orders before filing in any new court. Subscribe to your bar association’s ethics updates and practice area alerts.
The Bottom Line for Legal AI Adoption
The regulatory environment for AI in legal practice is evolving, but the core principle is clear: existing professional responsibility rules apply to AI use. Competence, supervision, confidentiality, and candor obligations don’t disappear because AI is involved — they extend into the AI workflow.
The good news: the purpose-built legal AI tools reviewed on this site are designed with attorney professional responsibility in mind. They maintain confidentiality, are designed to support (not replace) attorney judgment, and help reduce — not increase — the risk of errors in legal work.
The risk comes from using general-purpose AI tools without appropriate supervision, from submitting AI outputs without verification, and from failing to understand the limitations of the tools you use.
Understand your tools, review your outputs, and maintain your professional obligations. The AI serves you — not the other way around.
legalaireviews.net/ covers AI tools for legal professionals. This article provides general information about professional responsibility guidance and is not legal advice. Consult your state bar for jurisdiction-specific ethical guidance.
Related Articles
How Law Firms Are Responding to AI Ethics Requirements
The practical response from law firms ranges from proactive to reactive:
Firms with formal AI policies have typically developed written policies covering: (1) which AI tools are approved for use, (2) what types of matter data can be submitted to which tools, (3) how AI outputs must be supervised and reviewed before client or court delivery, and (4) how attorneys should disclose AI use to clients when appropriate. AmLaw 200 firms have largely formalized these policies; smaller firms are catching up.
Firms without formal policies are relying on individual attorneys to navigate AI use under existing professional conduct rules. This works for attorneys who understand the rules — and creates significant risk for those who don’t.
The most common error continues to be citation verification failure — attorneys submitting AI-generated research to courts without independently confirming that the cited cases exist and say what the AI claims they say. Courts have shown limited patience for this. After the Mata v. Avianca sanctions in 2023, similar cases have continued to appear in federal and state courts.
AI Ethics: Emerging Issues to Watch
Client consent for AI use. While ABA Opinion 512 doesn’t require disclosure in all cases, several state bars are moving toward explicit client consent requirements for certain types of AI use — particularly when client confidential information is submitted to third-party AI systems. Attorneys should monitor state bar guidance developments and consider proactive consent language in engagement letters.
AI in litigation: discovery of AI use. Courts are beginning to ask whether AI tool usage in discovery processes, document review, and case preparation is discoverable. This remains unsettled, but attorneys who use AI in document review workflows should consider how they would respond to discovery requests about their review methodology.
Unauthorized practice concerns for AI tools. Some bar associations have raised questions about whether AI tools that provide legal analysis constitute unauthorized practice of law when accessed directly by non-lawyers. Law firms should be aware that facilitating AI access to clients without appropriate attorney oversight could raise similar questions.
Frequently Asked Questions
Does using AI tools mean I need to disclose this to clients? ABA Formal Opinion 512 does not require disclosure in all circumstances, but recommends transparency — particularly when AI significantly affects billing, when clients have concerns about AI, or when AI is used on sensitive matters. Including a brief AI use policy in your engagement letter is increasingly common practice and is advisable.
Can law firms charge for AI tool costs on client matters? This is an area of active development. Some firms are treating AI tool usage like other technology costs — either absorbed as overhead or passed through as a billed disbursement. The ABA’s guidance on fees (Rule 1.5) suggests that billing clients for time that would have been required without AI, when AI reduced that time significantly, may create issues. Transparent communication with clients about how AI affects billing is the defensible approach.
What happens if a client objects to AI use on their matter? If a client objects, the attorney must respect that preference. The practical response is to ensure engagement letters address AI use upfront so that client expectations are set before work begins rather than after a disagreement arises.