A lawyer files a brief. The judge flags six citations that don’t exist. The attorney blames the AI. Courts and disciplinary bodies, however, have been clear: the responsibility stays with the lawyer. That scenario has already played out in federal courts across the country, producing sanctions, disciplinary referrals, and in some cases, suspensions. It’s no longer a hypothetical risk. It’s the professional environment lawyers navigate right now, and why the conversation about lawyer use of AI ethics has moved from theoretical to urgent.
In July 2024, the ABA issued Formal Opinion 512, its first ethics opinion focused squarely on generative AI. The opinion doesn’t create new rules. Instead, it maps existing Model Rules onto AI use in ways every practicing attorney needs to understand before opening another chatbot window. The obligations it describes apply to work you do yourself, work you delegate to AI tools, and work done by any third party you authorize to act on your behalf, including your marketing agency. This guide explains lawyer use of AI ethics and what attorneys must do to remain compliant under those Model Rules and ABA Formal Opinion 512.
Thrive Business Marketing works exclusively with law firms, which means we operate under the same scrutiny this guide covers. By the end of this article, you’ll know your core obligations, where state bars diverge from the ABA baseline, what disciplinary failures look like in practice, and how to run a quick compliance check before connecting any AI tool to your client matters.
What ABA Formal Opinion 512 Establishes as Your Baseline
Formal Opinion 512 isn’t binding law on its own, but it’s the closest thing to a national standard lawyers currently have for AI ethics. Many state bars developing their own guidance are working from the same foundation, which makes understanding this opinion the starting point for any AI compliance review. While not every state bar has explicitly cited the ABA opinion, it remains the dominant reference point across jurisdictions.
The opinion doesn’t treat AI as a new category of ethics problem. It treats AI as a tool that triggers rules you already know. Rule 1.1 requires competence, which now means understanding how the AI tool you’re using works well enough to catch its errors. Rule 1.4 means communicating with clients when AI use affects how their matter is being handled. Rule 1.5 means not billing clients for time AI saved you, some bar guidance has specifically flagged billing a client for the hours spent learning a new tool as problematic, though treatment varies by jurisdiction.
Rule 1.6 means client information you enter into an AI system must be protected with the same rigor as any other confidential disclosure. Rules 3.1 and 3.3 mean filings must be accurate, period, regardless of where the first draft came from. Rules 5.1 and 5.3 mean AI output is supervised work product, and the supervising attorney remains responsible for everything in it.
The opinion’s central principle is straightforward: AI can be a legitimate starting point for legal work, but no lawyer can delegate professional judgment to the tool. The output carries your name. The responsibility is yours.
Lawyer Use of AI Ethics: The Four Core Obligations
Competence under Rule 1.1 now means more than knowing how to log into a platform. It means understanding what the tool does with your inputs, where it gets its information, how it generates output, and what kinds of errors it’s likely to produce. Supervision under Rules 5.1 and 5.3 extends to AI the same way it extends to any nonlawyer on your team: you’re responsible for reviewing the work, catching the errors, and ensuring the final product meets professional standards. In practice, that means reading every AI output critically and verifying every citation against a primary source. Some state bars, Texas, for example, specifically recommend documenting that verification; treating it as a universal best practice regardless of your jurisdiction is a sound approach.
Attorney confidentiality and AI present a specific workflow requirement under Rule 1.6 before you ever type a client’s name into an AI prompt. Client information entered into an AI tool may be stored, processed, or used to train the model, and any of those outcomes can create a Rule 1.6 problem if you haven’t vetted the vendor and secured appropriate protections first. Some enterprise agreements prohibit training on client content; securing one before connecting client data is essential to your compliance posture. Vet the vendor first, not after a problem surfaces.
Disclosure under Rule 1.4 and fee reasonableness under Rule 1.5 work together. If AI use affects how the client’s matter is handled in ways that matter to their decisions, they need to know. Billing for time AI eliminated is unreasonable under Rule 1.5, and charging a client for time spent learning a new tool at their expense has been flagged as problematic by several bar authorities, though jurisdictions differ on the precise standard. Transparent billing in an AI-assisted practice means charging for the attorney judgment you exercised, not the time the machine saved you.
How State Bar Guidance Diverges from the ABA Baseline
The ABA opinion gives you a framework, but your state bar controls your obligations. California’s guidance is among the most demanding in the country. California State Bar Formal Opinion 2025-1 addresses affirmative client disclosure when AI is used to draft, summarize, or analyze material sent to a client or filed in a proceeding. The opinion is fact-specific, and California’s approach leans toward requiring explicit disclosure rather than allowing it to be buried in generic engagement language, consult the opinion directly for your specific circumstances. A boilerplate “technology-assisted services” clause in your engagement letter is unlikely to satisfy that standard.
Florida takes a different approach. Florida Bar Ethics Opinion 24-1 requires disclosure when AI use would affect a client’s decision about the representation, framing the issue as a materiality and informed-consent question rather than a task-based rule. Texas requires documented evidence that a lawyer reviewed AI output for accuracy and the absence of hallucinated authority before relying on it. New York City requires disclosure only when AI use is “substantial” in producing the final work product, a threshold the opinion itself leaves undefined.
Illinois, New Jersey, Colorado, Virginia, and more than a dozen other jurisdictions have issued their own guidance, and the differences matter. The ABA baseline is persuasive, not controlling. Your state bar’s position is what determines whether you’re compliant, and checking it is a non-negotiable step before building any AI workflow into your practice.
What Disciplinary Cases Reveal About How AI Failures Happen
The pattern documented across recent cases is consistent. A lawyer uses AI-generated research, files citations that don’t exist, and faces sanctions when a judge or opposing counsel discovers the fabrication. Steven Schwartz and Peter LoDuca were fined in Manhattan federal court after submitting ChatGPT-generated citations in 2023. Mike Sethi and William Rounds were sanctioned by the Ninth Circuit in 2026 for briefs containing fictitious cases and suspended from practicing before that court for six months. Kathleen Wilson and Shaun Hunter Rideway received two-year suspensions in Mississippi after their filings included fabricated case law. A Canadian lawyer was suspended for six months after relying on ChatGPT-generated fake case law in court submissions.
Courts and disciplinary bodies have been consistent in one finding: using AI is not the misconduct. Submitting unverified AI output is. Every lawyer sanctioned in the recent record failed to independently verify what the tool produced before relying on it. Some doubled down when courts raised questions, which transformed a competence problem into a candor problem under Rule 3.3. The lesson isn’t to avoid AI. It’s that your duty of accuracy is nondelegable, regardless of where the first draft came from.
Practical Checklist for Lawyer Use of AI Ethics in Your Daily Workflow
Vendor Due Diligence: Questions to Ask Before You Sign Up
Before connecting any AI tool to client matters, complete a brief vendor intake, in writing, before you sign anything. Ask the vendor these questions:
- Does the vendor use your data for model training or product improvement?
- What are the default data retention periods for prompts, outputs, and logs?
- Who at the vendor can access your content, and under what controls?
- Is data encrypted at rest and in transit?
- What happens to your data on contract termination?
- Will the vendor indemnify you for breaches attributable to their system?
If the vendor won’t answer these questions in writing, that’s your answer. This isn’t a legal review, it’s a 30-minute intake process that’s essential to AI risk management for law firms before any tool touches client information.
Pre-Filing Verification Steps
Before you file or send any AI-assisted work product, run through these verification steps:
- Check every case citation against Westlaw, Lexis, or a reliable primary source.
- Confirm that quoted language appears verbatim in the cited source.
- Verify that statutes cited reflect current law.
- Review the overall analysis for internal consistency errors an AI might miss.
This process fulfills the supervision duty under Rules 5.1 and 5.3 and protects you from the pattern of sanctioned conduct the disciplinary record documents.
Client Disclosure Language
For client disclosure, this language gives you a workable starting point you can adapt: “Our firm uses AI-assisted drafting tools to improve efficiency in preparing documents and research summaries. All AI-generated content is reviewed and verified by a licensed attorney before use. If you have questions about how AI tools are used in your matter, please let us know.”
For California-governed matters, California’s guidance leans toward requiring more specific disclosure and a written agreement covering the type of AI use and client consent, consult Formal Opinion 2025-1 directly for the details applicable to your situation.
Why Your Marketing Partner’s AI Ethics Matter as Much as Your Own
When a law firm authorizes a third-party vendor to use AI to create content, advertising, or messaging on its behalf, the firm’s name goes out with that output. Attorney advertising rules apply to AI-generated marketing content the same way they apply to everything else the firm publishes. If a marketing agency uses AI to write your firm’s website copy, blog articles, or ad text without verifying accuracy or flagging potential compliance issues, the professional and reputational exposure lands on you.
Thrive Business Marketing works exclusively with law firms, which means the team understands that attorney advertising rules, accuracy obligations, and attorney confidentiality in AI-assisted work don’t stop at the office door. When Thrive uses AI tools to support content creation, campaign strategy, or SEO for attorney clients, it applies human review and verification of all AI-generated output, along with strict data handling protocols that protect client information and full transparency about how AI is used in each marketing program.
For lawyers evaluating any marketing partner, the questions mirror those in this guide’s vendor checklist: who reviews the AI output, what data do they process, and how do they handle confidential information? Any agency that can’t answer those questions clearly isn’t a safe partner for a law firm.
Build the Habit Before You Need It
ABA Formal Opinion 512 provides the model-rule framework. Your state bar’s guidance controls your actual obligations. And every disciplinary case in the recent record traces back to unverified reliance on AI output. Building strong habits around lawyer use of AI ethics is the practical response, not avoiding these tools, but developing a short, repeatable verification routine that treats every AI output as a first draft requiring professional review before it leaves your hands.
The regulatory picture is moving fast. State bars are issuing new guidance frequently, courts are building their own AI disclosure requirements, and the bar for competence under Rule 1.1 continues to rise as these tools become more widely used. Lawyers who build ethical AI habits now are in a significantly stronger position than those who wait for a disciplinary proceeding to prompt the review.
For law firms that want a marketing partner who takes these obligations seriously, Thrive Business Marketing brings that same standard to legal marketing. Visit Thrive Business Marketing to learn how a specialized legal marketing agency can help your firm grow without cutting corners on compliance.