First AI Class

A Step-by-Step Guide to Using AI Safely as a Lawyer

Reviewed by Stephen J. Ronan, MD

Last verified:


You've seen the headlines about AI rewriting legal research and drafting documents in seconds. Maybe you've wondered if you should start using it - but you're not sure where to begin, or if it's even allowed under your ethics rules. That's exactly why you're here.

You're a practicing lawyer who wants to understand how to start using AI safely, without risking ethics complaints, client trust, or accuracy. This isn't about replacing your skills. It's about adding one careful tool to your practice. Let's walk through how.

Why You Should Care About AI Safety First

AI tools for lawyers aren't magic. They're trained on data, and sometimes that data is incomplete, outdated, or wrong. An AI might generate a citation to a case that doesn't exist or misstate a statute. If you rely on that without checking, you could mislead a client or submit a defective filing.

Client confidentiality is the other big worry. Many free AI tools process your input through public servers. If you paste client details into one of those, you may expose sensitive information. Even if the tool claims "anonymization," there's no guarantee the data stays private.

State bar rules are evolving, too. Several bar associations have now issued guidance requiring lawyers to verify AI output under their existing competence duties (often tied to Rule 1.1). Using AI doesn't absolve you of your duty to ensure accuracy. You don't need to be an AI expert - but you do need to understand how these tools work and where they fail. A good starting point is our guide to client confidentiality when using AI.

Three Free AI Tools That Actually Work

You don't need to buy expensive software to get comfortable. Here are three free options that fit ethical and practical needs for legal work:

  1. Google Gemini is a reasonable starting point for basic case law scoping. It can't replace Westlaw or Lexis, but it helps you quickly identify potential cases to cross-check. If you're drafting a motion to dismiss, you might ask, "What are recent federal court decisions about Rule 12(b)(6) motions?" The results give you keywords and jurisdictional hints - but you still verify in your subscription database.

  2. Spellbook offers contract review features useful for transactional lawyers. You can use it to flag unusual clauses in a draft NDA or services agreement. It logs edits and outputs, which helps with documentation.

  3. ChatPDF lets you upload a PDF - a court opinion, a contract, a deposition transcript - and ask questions about its contents. This is helpful for parsing dense documents. You might ask, "What's the holding of this case regarding standing?" It will summarize, but you still confirm the summary matches the original.

Why start with free tools? They let you experiment without financial risk and help you build a feel for AI's strengths and limits. Once you're comfortable, you can look at paid tools - our best AI tools for lawyers guide covers options like Harvey and Lexis+ AI in detail.

A 2-Minute Workflow to Stay Ethical

Here's a simple three-step process that keeps you compliant and effective on any AI-assisted task.

1. Input only non-confidential information. Before using any AI tool, redact client names, case numbers, and sensitive facts. If you're drafting a contract, input only the type of agreement and key terms: "Create a non-disclosure agreement for a tech startup with a $5M revenue cap." That minimizes data exposure.

2. Verify every citation and fact. If you ask AI to suggest cases for a brief, don't cite them blindly. Cross-check each case in Westlaw or Lexis. If the AI says "Case XYZ held that...", look it up and confirm the holding matches. This step is non-negotiable under bar ethics.

3. Document your process. Keep a log of what you asked, what the tool returned, and how you verified it. If a client asks how you found a case, or a court questions a citation, this record shows you acted competently. A simple spreadsheet with dates, queries, and verification notes is enough.

This workflow isn't about distrust. It's about accountability. You're not just using AI; you're making sure it works with you, not against you. Doctors are building parallel workflows for the same reason - if you're curious how clinicians are approaching it, see how doctors start using AI safely.

Common Mistakes Beginners Make

A few pitfalls trip up lawyers new to AI.

Using AI for privileged communications. If you draft a strategy memo in a consumer AI tool, that communication could be exposed if the tool's data is subpoenaed or breached. Assume anything you input to a free tool may become public.

Ignoring state bar guidance. Several state bars have issued opinions requiring "reasonable care" when using AI, including verifying outputs. If you skip that step and make an error, the bar can question your competence. Check your state's rules - they're not optional.

Using unverified case law in filings. Multiple federal courts have now sanctioned lawyers for citing AI-hallucinated cases. You can avoid this by treating AI like a first-year associate: it drafts, you confirm.

Overlooking data privacy settings. Enterprise versions of AI tools often offer encryption and data retention controls; free versions usually don't. Ask before you use a tool: Does this store my input? Can third parties access it? When in doubt, don't input it.

Next Steps: Your AI Readiness Checklist

Here's how to build your own readiness plan.

First, take the 2-minute quiz on our AI for lawyers hub. It identifies your biggest risks and gaps, and it's designed for lawyers who want to move forward confidently - no jargon, no pressure.

Next, join a local AI workshop. Many state and local bar associations now offer training on ethical AI use. These are good for hands-on learning and peer support, and many qualify for CLE credit.

Finally, start with one task. Pick a routine task you spend 10-30 minutes on each week - a standard motion, a client intake summary, a contract review checklist. Use AI to streamline it, and follow your verification workflow every time. Once that task is running smoothly, add another. A solo practice attorney might test Spellbook on contract redlines for three standard NDAs this month. Document each verification step in a log, then reflect: Did this save time? Did I miss anything? Would I do it again next month?

One Small Step to Start

The goal isn't to become an AI expert. It's to add one tool to your practice without compromising ethics or quality. You've already taken the first step by reading this. Now take the quiz to see where you stand - every lawyer using AI safely today started exactly where you are.

Frequently asked questions

What AI tools are safe for lawyers to use?
Start with tools that don't retain your data, like enterprise versions of ChatGPT or Spellbook. Free tools are fine for non-confidential tasks only.
How do I verify AI-generated legal research?
Treat every citation as a lead, not a source. Look up each case in Westlaw or Lexis and confirm the holding before you use it.
Can I use free AI tools for client work?
Only if you strip out confidential details first. Redact names, case numbers, and facts that could identify a client.
Do I need to tell the court I used AI?
Some federal judges now require disclosure. Check your jurisdiction's standing orders and your state bar guidance before filing.
What if AI gives me bad legal advice?
You're responsible under Rule 1.1 competence duties. Verify every output, document your process, and never cite unchecked AI content.
How do I protect client data when using AI?
Use enterprise tools with data retention controls, redact identifiers before input, and confirm the tool's privacy settings in writing.

Related reading