Can Lawyers Use ChatGPT? What ABA Formal Opinion 512 Means in Practice
Contents
With conditions. ABA Formal Opinion 512, released on July 29, 2024, is the American Bar Association's first ethical opinion on the use of artificial intelligence for legal matters. In this opinion, the ABA lays out the existing ethical rules that apply when lawyers utilize the AI tool in their legal practice. This article aims to provide a summary of ABA Formal Opinion 512 for lawyers who use AI to review legal contracts and will provide a helpful vendor checklist for you to utilize this week.
Can lawyers use ChatGPT under ABA Formal Opinion 512?
The short answer is yes. ABA Formal Opinion 512 does not ban generative AI. It says you may use it if you still meet your duties of competence, confidentiality, communication, supervision and reasonable fees. Three points matter most:
- Verify every output yourself You remain responsible for the work, whatever tool produced the first draft
- Protect client information Do not enter confidential details into a tool that may store them or train on them unless the client gives informed consent
- Bill honestly If AI cuts a task from five hours to one, you bill for the time you spent
Two caveats. ABA Formal Opinion 512 is guidance, not binding law. Your state's rules of professional conduct and your state bar's AI guidance control and several states have published their own. This article is general information, not legal advice.
What ABA Formal Opinion 512 says
ABA Formal Opinion 512 walks through the Model Rules that apply when you use generative AI. Here is each duty and what it asks of you.
Competence Rule 1.1
You need a reasonable understanding of what the tool can and cannot do and you need to keep that understanding current. You do not have to become an engineer. ABA Formal Opinion 512 also says you must review the output and the level of review depends on the task. Skimming an AI summary of your own meeting notes needs less checking than a brief that cites case law.
The standard example is Mata v. Avianca. In 2023, a New York federal court sanctioned two lawyers $5,000 after they filed a brief with case citations that ChatGPT had invented. The lawyers never checked the cases.
Confidentiality Rule 1.6
When you use artificial intelligence for legal purposes, you have a responsibility under the ABA's Confidentiality Rule 1.6 to ensure that the client’s confidential information is not disclosed to any external third party without the client’s consent. Additionally, AI tools that learn from your inputs pose a risk to your clients because one client’s confidential information could potentially be disclosed in the AI’s outputs to another party who is using the tool. Therefore, under Formal Opinion 512, you generally have to obtain the client’s informed consent before utilizing these types of AI tools to review contracts or other legal information.
Communication Rule 1.4
You must keep clients reasonably informed. That includes telling them about AI use when it affects the representation for example when you need consent to enter their data, when they ask or when AI use changes how you charge.
Supervision Rules 5.1 and 5.3
Partners and managers must set clear policies for AI use and train the people who use it. The same duty covers nonlawyer staff. Treat AI output the way you treat work from a junior colleague: someone qualified checks it before it goes out.
Fees Rule 1.5
If you are billing clients for legal services on an hourly basis, the AI tool can reduce the time that it takes to perform a contract review. For example, if a contract took a lawyer 5 hours to review, using artificial intelligence to help with the contract review only takes a lawyer 20 minutes of input into the AI tool and then another 40 minutes of output checking of the AI-generated first pass of the contract. In this scenario, you can only bill the client for 1 hour of legal service rather than the 5 hours it would have taken you to perform the review on your own. You can continue to charge clients a flat fee for services, but these fees must still be reasonable to the clients. Additionally, you are unable to pass on general overhead charges to clients for utilizing artificial intelligence tools unless they agree in advance to these types of charges. Finally, you are not able to bill clients for the time that you take to learn how to utilize any of the artificial intelligence tools you begin to use in your legal practice.
Candor and meritorious claims (Rules 3.1 and 3.3)
You cannot submit false statements to a court and that includes false statements an AI tool wrote for you. Check citations, check quotes and follow any court rules on AI disclosure.
What ABA Formal Opinion 512 means for contract review
Contract review is one of the most common legal uses of AI and ABA Formal Opinion 512 applies to it directly. A contract carries client names, prices, deal terms and sometimes personal data.
1 Choose the tool first Run the vendor questions in the next section before any client document goes in.
2 Decide whether you need consent If the tool may retain or train on inputs, get specific informed consent from the client in writing
3 Limit what you enter. Remove names, account numbers and other details the review does not need
4 Check every flagged clause against the source text Open the contract, read the clause and confirm the AI described it correctly
5 Verify every legal citation Confirm that the case or statute exists and says what the tool claims
6 Keep a record Note which tool you used, what you entered and who reviewed the output
7 Bill actual time Record the minutes you spent, not the hours the task used to take
8 Tell your team the rules Put your AI policy in writing and train everyone who touches client files
Confidentiality and tool selection: vendor due-diligence questions
ABA Formal Opinion 512 expects you to read a tool's terms of use and privacy policy before you rely on it. Free consumer chatbots often allow the provider to keep and use your inputs. Ask any vendor these questions and get the answers in writing:
• Does the vendor or its model provider train on your inputs or outputs?
• Is there a zero data retention agreement with the underlying model provider and can the vendor show it?
• How long are documents, prompts and outputs stored and where?
• Which subprocessors handle your data?
• Is data encrypted in transit and at rest and does the vendor hold a current SOC 2 Type II report or ISO 27001 certificate?
• Who at the vendor can see your documents and is access logged?
• How quickly will the vendor tell you about a breach? Ask for 24 to 48 hours.
• Can you delete your data on request and how fast does deletion happen?
If the answers are vague, treat the tool as one that may keep your data and plan for client consent or avoid putting client documents into it.
How iDeal Legal AI supports your ABA 512 duties
iDeal Legal Ai is built around the duties in ABA Formal Opinion 512. Documents are processed under zero data retention terms, so inputs and outputs are not stored by the model providers after processing and are not used to train models. Every flagged clause links back to the exact text in your contract, so you can verify it in seconds. A lawyer reviews and approves the results before anything reaches a client.
That design supports your duty of competence, because you can check each finding. It supports your duty of confidentiality, because the data does not persist. It supports your duty of supervision, because the review step is built into the workflow.
ABA Formal Opinion 512 FAQ
Can lawyers use ChatGPT?
Yes. ABA Formal Opinion 512 allows lawyers to use generative AI tools such as ChatGPT if they meet their ethical duties. You must understand the tool, verify its output, protect client confidentiality, supervise its use and bill fairly. For client-confidential material, a consumer version of ChatGPT that may retain or train on your inputs usually calls for informed client consent or a different tool.
What is ABA Formal Opinion 512?
ABA Formal Opinion 512 is the American Bar Association's ethics opinion on generative AI tools, issued on July 29, 2024. It explains how the Model Rules on competence, confidentiality, communication, supervision, candor and fees apply when lawyers use AI.
Is ABA Formal Opinion 512 binding?
No. It is persuasive guidance. Each state adopts and interprets its own rules of professional conduct and your state bar's rules and opinions control. Many state bars follow the ABA's approach, so the opinion is a strong indicator of what regulators expect.
Where can I find the ABA Formal Opinion 512 PDF?
The ABA publishes the full text on its website under the Standing Committee on Ethics and Professional Responsibility. Read the original opinion, not a summary, before you write your firm's AI policy.
What does ABA Formal Opinion 512 say about confidentiality?
It says you must evaluate the risk of disclosure before entering client information into an AI tool and that you generally need informed consent when the tool may expose that information to others. General boilerplate in an engagement letter does not satisfy the requirement.
Do I have to tell clients I use AI?
Not in every case. You must tell clients when AI use affects the representation, for example when you need consent to enter their data, when they ask about it or when it changes your fees. Many firms disclose AI use in the engagement letter as a matter of practice.
Can I bill clients for time AI saves me?
No. Under ABA Formal Opinion 512, hourly billing reflects the time you actually spend. If AI reduces a five-hour task to one hour, you bill one hour. A reasonable flat fee is allowed.
What are the main AI ethics rules for lawyers?
AI ethics for lawyers come from the existing Model Rules: competence (1.1), communication (1.4), fees (1.5), confidentiality (1.6), candor to the court (3.3) and supervision (5.1 and 5.3). ABA Formal Opinion 512 explains how each rule applies to generative AI.
Source
ABA Standing Committee on Ethics and Professional Responsibility, Formal Opinion 512, Generative Artificial Intelligence Tools (July 29, 2024).