When I talk to lawyers about AI, the first question is almost never “how do I use it?” It’s “am I allowed to use it?” That instinct — to check the ethical guardrails first — is exactly right.

Using AI in legal practice isn’t just a technology question. It’s an ethics question. Bar associations across the country are issuing guidance (over 30 state bars have weighed in as of early 2026), and courts are imposing sanctions on lawyers who use AI carelessly. The rules are still evolving, but the direction is clear.

Here’s what you need to know right now.

The Competence Obligation

Model Rule 1.1 requires lawyers to provide competent representation. Multiple bar associations have interpreted this to include understanding the technology you use — including AI.

This means:

• You need to understand what AI can and can’t do
• You need to know that AI can fabricate case citations
• You need to verify AI output before submitting it
• “The AI wrote it” is not a defense for errors

Several lawyers have already been sanctioned for submitting AI-generated briefs with fabricated citations. The courts were not sympathetic.

Confidentiality Concerns

Model Rule 1.6 requires protecting client information. When you paste client details into an AI tool, you need to consider:

• Where does the data go? Consumer AI tools (free ChatGPT, free Claude) may use your input for training
• Who can access it? Cloud-based tools store data on third-party servers
• Is it encrypted? Not all AI tools encrypt data in transit and at rest

Safe Practices:

• Use enterprise AI tools with data privacy agreements (ChatGPT Enterprise, Claude for Business, CoCounsel)
• Never paste client names, case numbers, or identifying details into consumer AI tools
• Anonymize information before using AI: replace names with “Party A” and “Party B”
• Check your firm’s AI policy — if you don’t have one, create one

Disclosure to Clients

Should you tell clients you’re using AI? The answer is increasingly yes:

• Some jurisdictions now require disclosure of AI use in legal work
• Several courts require attorneys to certify whether AI was used in filings
• Even where not required, transparency builds trust

A simple disclosure works: “Our firm uses AI tools to assist with drafting and research. All AI-generated content is reviewed and verified by a licensed attorney before use.”

Disclosure to Courts

Multiple federal courts now require AI disclosure in filings. The typical requirement:

• Certify whether AI was used in preparing the filing
• If AI was used, confirm that all citations have been verified
• Some courts require identifying which AI tool was used

Check your jurisdiction’s local rules. This is changing rapidly.

Billing Ethics

Can you bill for AI-assisted work? Yes, but with considerations:

• Don’t bill for AI time — if AI drafts a letter in 30 seconds, you can’t bill for the hour it would have taken manually
• Bill for your review time — the time you spend reviewing, editing, and applying judgment is billable
• Be transparent — if a client asks how you produced work so quickly, be honest
• Pass savings to clients — AI should reduce client costs, not inflate your margins

Creating a Firm AI Policy

Every firm needs an AI policy. At minimum, it should cover:

1. Approved tools — which AI tools are authorized for use
2. Data handling — what information can and cannot be entered into AI tools
3. Verification requirements — all AI output must be verified before use
4. Disclosure obligations — when and how to disclose AI use
5. Training requirements — who needs training and how often
6. Incident reporting — what to do if AI causes an error

The Bottom Line

AI is a powerful tool for lawyers. But power without responsibility leads to sanctions, malpractice claims, and damaged reputations. The lawyers who use AI ethically — with verification, confidentiality protections, and transparency — will thrive. The ones who use it carelessly will learn expensive lessons.

Stay informed. Your bar association’s ethics opinions on AI are worth reading. They’re being updated regularly.