Ethics Safe AI Use for Law Firms
ABA Formal Opinion 512:
What It Requires
Issued July 2024. Applies to every ABA member in the country. Twelve specific obligations. Most firms have addressed none of them in writing.
What Is ABA Formal Opinion 512?
ABA Formal Opinion 512, Generative Artificial Intelligence Tools, was issued by the ABA Standing Committee on Ethics and Professional Responsibility in July 2024. It is the first comprehensive guidance from the ABA on how the Model Rules of Professional Conduct apply to the use of generative AI tools in legal practice.
The Opinion does not create new rules. It maps existing rules — competence, confidentiality, supervision, candor, and others — onto the specific risk profile of generative AI. The result is a set of twelve concrete obligations that apply to any attorney who uses, or whose staff uses, AI tools in client matters.
Opinion 512 is not binding law. But it reflects the national consensus on what competent, ethical AI use looks like — and it is the document courts, bar disciplinary panels, and malpractice carriers will reference when something goes wrong.
The Opinion issued in July 2024. Mata v. Avianca sanctions were imposed in June 2023. Courts were already ahead of the ethics authorities. The judicial record has continued to grow since.
The Twelve Obligations
Opinion 512 maps to six Model Rules. Each rule generates specific, actionable obligations. Here is what the Opinion actually requires.
- Understand the capabilities and limitations of each AI tool you use.
- Know how the tool was trained and what it is likely to get wrong.
- Maintain competence as AI tools evolve — this is an ongoing obligation, not a one-time assessment.
- Client confidential information may not be input into AI tools without adequate confidentiality protections.
- Free-tier AI tools almost universally use inputs for model training — they are off-limits for client matter work.
- Vendor agreements must include a Business Associate Agreement (BAA) for any tool that processes protected health information.
- Supervising partners are responsible for AI use by subordinate attorneys and non-lawyer staff.
- A written AI policy must exist. Verbal instructions are not sufficient.
- The policy must cover which tools are approved, for which tasks, and what verification is required before filing or sending.
- Attorneys may need to disclose AI use to clients, depending on engagement letter terms and the nature of the matter.
- Engagement letters should address AI use — what is permitted, what is prohibited, and whether client consent is required.
- An attorney who submits AI-generated content to a tribunal has a duty to verify that it is accurate.
- "The AI produced it" is not a defense to a Rule 3.3 violation. It is the beginning of the inquiry.
- Citation verification must occur before every AI-assisted filing — not as a spot check, but as a documented step.
- AI-generated efficiency gains must be reflected in billing practices.
- Billing a client at the same rate for AI-assisted work that took a fraction of the time may implicate Rule 1.5.
What It Means for a PI Firm
Opinion 512 was written for all attorneys. But personal injury firms face a specific risk profile that makes compliance both more urgent and more concrete.
The volume problem. A PI firm with 400 active files and three paralegals who all have ChatGPT on their phones is not facing one AI risk — it is facing 400 simultaneous AI risks. Every demand letter, every medical record summary, every intake form that touches an AI tool is a potential Rule 1.6 violation if the tool lacks adequate data protections, and a potential Rule 3.3 violation if the output goes out unverified.
The contingency fee problem. A PI firm that loses a contingency fee because of an AI-driven sanctions order does not lose a billable hour. It loses the entire fee on that case — potentially $50,000, $200,000, or more. The asymmetry between the cost of a written AI policy and the cost of one sanctions order is enormous.
The supervision problem. Under Rules 5.1 and 5.3, the managing partner is personally exposed when a subordinate uses an AI tool without authorization or without a verification protocol. The Opinion makes clear that "I didn't know they were using it" is not a defense if a reasonable supervisory system would have caught it.
More than 500 cases involving lawyer AI misconduct had been reported in U.S. courts as of early 2026. In every one, the same fact was present: no documented verification process was in place when the error occurred.
What Monday Morning Looks Like
Opinion 512 tells you what the rules require. It does not tell you what to do with six attorneys, 400 active files, and three paralegals who all have ChatGPT on their phones.
That gap is what Ethics Safe AI Use for Law Firms is designed to close. The playbook maps every obligation in Opinion 512 to a specific operating procedure, checklist item, or policy clause — built for how a PI firm actually runs.
The Protocols That Close the Gap
Ethics Safe AI Use for Law Firms maps Opinion 512's twelve obligations to specific procedures for intake, demand letters, pleadings, medical records, client communication, and advertising — with a 47-point audit checklist and ready-to-use disclosure language for N.D. Cal. and C.D. Cal. filings.
Firm license (up to 10 attorneys): $199