Last reviewed
The short answer. Lawyers can use ChatGPT and Claude for drafting, summaries of public law, and plain-English client explanations. ABA Formal Opinion 512 says informed consent is required before information about a representation enters a self-learning tool, and boilerplate consent is not enough. Every output still needs the lawyer’s review.
Where ChatGPT helps a practice
Draft letters, routine motions and client updates from placeholder facts; summarize a published opinion, statute or regulator guidance; explain a process in plain English, then check it before sending.
What Opinion 512 asks of you
The opinion applies competence, confidentiality, communication, reasonable fees, candor, and supervision. Lawyers must understand a tool's terms and privacy policy, or consult someone who does. Hourly billers charge time actually spent and may not bill a client for general tool training.
State rules govern
State rules and opinions control. Florida Ethics Opinion 24-1 recommends informed consent before confidential information goes to a third-party AI program, and California publishes practical generative-AI guidance.
Verify every citation
Mata v. Avianca imposed a $5,000 sanction after lawyers filed invented opinions. Read every case and quotation in the reporter or record before filing.
The habit that makes it safe
Replace client, opposing-party and witness names with [CLIENT] and [OPPOSING_PARTY], keep only necessary facts, preserve labels, and use a firm-administered account.
What lawyers use ChatGPT for
These uses begin with public, generic, or synthetic material and end with professional review.
- Drafting an outline from a hypothetical fact pattern
- Creating a checklist from public court rules for a lawyer to verify
- Rewriting generic client education material in plain language
- Brainstorming research terms without client or matter facts
- Editing an internal template after confidential fields are removed
Client data that stays out of prompts
Use placeholders or synthetic records when the task needs structure. Keep the following data out of consumer and unapproved AI tools:
- Client, witness, employee, and opposing-party identities and contact details
- Privileged and confidential communications
- Matter facts, contracts, discovery, evidence, and investigation material
- Legal strategy, attorney work product, settlement positions, and offer terms
- Medical, employment, financial, government ID, and account data
- Passwords, access credentials, and API keys
Read Is ChatGPT safe for work? for provider data handling, training choices, retention, and encryption by plan type.
Five steps for safer use
- Start with policy. Define approved tools, prohibited data, review, reporting, and an owner in the AI acceptable use policy.
- Remove client context. Replace real records with synthetic examples and neutral placeholders before drafting the prompt.
- Use the approved account. Confirm the exact product, plan, settings, and connected features match company policy.
- Verify every result. Check facts, calculations, citations, professional requirements, and final language against authoritative sources.
- Report mistakes and exposure. Stop the workflow, preserve the relevant record, and follow the company reporting path.
The PII redaction guide shows how to define, test, and log a control before an LLM call. The AI glossary defines the gateway and guardrail terms.
Professional guidance
ABA Formal Opinion 512 says lawyers using generative AI must consider duties involving competence, confidentiality, client communication, supervision, candor, and reasonable fees. Its confidentiality analysis explains that lawyers must evaluate the tool and the information involved before entering information relating to a representation, including whether informed client consent is required. Follow the rules and opinions that govern the lawyer’s jurisdiction.
This guide provides operational information. Have counsel or the profession’s designated compliance owner review company policy and each approved use.
Frequently asked questions
Can lawyers use ChatGPT?
Yes. Lawyers remain responsible for competence, confidentiality, verification, client communication, supervision, candor, and fees under the rules that apply to their work.
Is ChatGPT confidential?
Do not assume a ChatGPT conversation has the confidentiality required for a legal matter. Evaluate the exact product, plan, settings, terms, and professional duties before use.
Can ChatGPT write legal documents?
It can assist with drafts in an approved workflow. A lawyer must verify the law, citations, facts, analysis, language, and final document.
Does attorney-client privilege apply to ChatGPT?
Privilege depends on the facts and governing law. Keep client and matter information out of prompts until the lawyer has completed the required confidentiality analysis.
Keep going
ChatGPT for accountants: use it without sending client data
ChatGPT for real estate agents: write faster, keep client data out
Using ChatGPT with client data
A practical next step
For regulated work
Sources
Sources read on 2026-10-07.
- ABA Model Rule 1.1https://www.americanbar.org/groups/professional_responsibility/publications/model_rules_of_professional_conduct/rule_1_1_competence/
- ABA Model Rule 1.6https://www.americanbar.org/groups/professional_responsibility/publications/model_rules_of_professional_conduct/rule_1_6_confidentiality_of_information/
- Florida Bar Ethics Opinion 24-1https://www-media.floridabar.org/uploads/2024/01/FL-Bar-Ethics-Op-24-1.pdf
- Mata v. Avianca sanctions orderhttps://www.courtlistener.com/docket/63107798/54/mata-v-avianca-inc/
- American Bar Association Formal Opinion 512https://www.americanbar.org/content/dam/aba/administrative/professional_responsibility/ethics-opinions/aba-formal-opinion-512.pdf
- SUPERWISE docs: PII redactionhttps://docs.superwise.ai/docs/pii-redaction
General information. It is not legal or tax advice. Reviewed 2026-10-07.