In June 2023, a New York attorney named Steven Schwartz made international headlines for submitting a legal brief containing six fabricated case citations: all generated by ChatGPT. He was sanctioned, fined $5,000, and became the cautionary tale every legal ethics CLE now references.

He won’t be the last. AI tools are powerful, but using them without understanding the ethical implications can genuinely end your career. Here’s what every practicing attorney needs to know.

Confidentiality (Rule 1.6)

The most immediate risk. When you paste client information into an AI tool, you may be disclosing confidential information to a third party.

The problem: Most consumer AI tools (ChatGPT, Claude, Gemini free tiers) may use your inputs for model training. That means client information could theoretically influence future outputs.

Safe practices:

• Never paste client names, case numbers, or identifying details into consumer AI tools
• Use anonymized or hypothetical versions of your facts
• Use enterprise AI tools with data processing agreements (CoCounsel, Harvey, Clio Duo)
• Check whether your tool’s terms of service include a no-training clause
• ChatGPT’s API and Team/Enterprise plans offer no-training guarantees. The free tier does not.

Template for anonymizing:

Instead of: “John Smith was injured at ABC Corp’s warehouse on January 15” Use: “A plaintiff was injured at the defendant employer’s facility. The key facts are…”

Competence (Rule 1.1)

You have a duty to understand the tools you use. “The AI told me” is not a defense for incorrect legal work.

What this means in practice:

• Understand how the AI tool generates output (probabilistic text generation, not legal reasoning)
• Know the tool’s limitations (hallucination, training data cutoffs, jurisdiction gaps)
• Verify every piece of AI output before relying on it
• Stay current on AI developments relevant to your practice

The Mata v. Avianca standard: An attorney was sanctioned for filing a brief with fabricated case citations generated by ChatGPT. The court found that the attorney failed in their duty of competence by not verifying the AI’s output.

Supervision (Rules 5.1, 5.3)

If your associates or staff use AI, you’re responsible for ensuring they use it properly.

Firm-level requirements:

• Establish a written AI use policy
• Train all attorneys and staff on approved tools and prohibited uses
• Require human review of all AI-generated work product
• Document AI use in work product where required

A basic AI policy should cover:

1. Which tools are approved for use
2. What types of information can and cannot be input
3. Review requirements before any AI output is used
4. Disclosure requirements to clients and courts
5. Who to contact with questions

Disclosure

A growing number of jurisdictions require disclosure of AI use in court filings.

Current landscape:

• Some federal judges require disclosure of AI use in all filings
• Several state bars have issued guidance requiring disclosure
• The trend is clearly toward more disclosure, not less

Safe approach: Disclose AI use in court filings proactively. A simple statement like “AI tools were used to assist in the preparation of this document. All legal research, analysis, and citations have been verified by the undersigned attorney” covers most requirements.

Billing

Can you bill for time spent using AI? The ethics are evolving.

Current consensus:

• You can bill for time spent reviewing and editing AI output
• You should not bill the same hours you would have spent doing the work manually if AI made it significantly faster
• Be transparent with clients about AI use if it materially affects billing
• Consider whether AI efficiency savings should be passed to the client

Client Consent

Should you tell clients you’re using AI?

Best practice: Yes. Include a brief AI disclosure in your engagement letter:

“Our firm may use artificial intelligence tools to assist with certain tasks such as research, document review, and drafting. All AI-assisted work is reviewed and approved by a licensed attorney. Client confidential information is only processed through enterprise-grade tools with appropriate data security measures.”

This builds trust and protects you if AI use becomes an issue later.

Building Your Firm’s AI Policy

Start with these five questions:

1. Which AI tools are approved? (List specific tools, not categories)
2. What data can be input? (Define what’s allowed and what’s prohibited)
3. What review is required? (Every output reviewed by a licensed attorney)
4. When must AI use be disclosed? (Court filings, client communications, both)
5. Who maintains the policy? (Designate an AI ethics point person)

Review the policy quarterly. AI capabilities and ethical guidance are changing fast.

The lawyers who thrive with AI won’t be the ones who use it most aggressively. They’ll be the ones who use it most responsibly.

Related reading: 7 Best AI Tools for Lawyers · AI for Legal Research · AI for Contract Review

🛠️ Try it yourself: Legal Document Drafter or Case Summary Generator: free, no signup needed.

FAQ

What are the biggest ethical risks of using AI in legal practice?

The three primary risks are: confidentiality breaches (inputting client data into tools that may use it for training), competence failures (relying on AI output without verification, such as filing hallucinated citations), and inadequate supervision (associates or staff using AI without proper guardrails or review requirements).

Do I need client consent to use AI on their matter?

Best practice is yes. Include a brief AI disclosure in your engagement letter explaining that your firm may use AI tools for certain tasks, that all work is reviewed by a licensed attorney, and that confidential information is only processed through enterprise-grade tools. This builds trust and protects you if AI use becomes an issue later.

Can I bill clients for work done with AI assistance?

You can bill for time spent reviewing and editing AI output, but you should not bill the same hours you would have spent manually if AI made the work significantly faster. Be transparent with clients about AI use when it materially affects billing. The ethical consensus is evolving toward passing some efficiency savings to clients.

What should a firm AI policy include?

A basic policy should cover: which specific tools are approved (not just categories), what data can and cannot be input, review requirements (every output reviewed by a licensed attorney), disclosure obligations for court filings and client communications, and a designated AI ethics point person. Review the policy quarterly as capabilities and guidance change rapidly.

How often should I verify AI-generated legal content?

Every single time: 100% verification rate. There is no threshold below which AI output can be trusted without checking. This applies to citations, case holdings, statutory references, and legal analysis. The standard is simple: would you stake your bar license on this being accurate? If not, verify it before relying on it.