Legal work is built on reading, writing and researching, which makes it one of the professions most affected by generative AI. Lawyers now use AI to summarize documents, draft first versions of contracts and briefs, review discovery and answer client intake questions. The productivity gains are real, but so are the risks: invented case citations, confidentiality breaches and billing questions have already led to court sanctions and new ethics guidance. This guide explains how law firms are using AI, the main categories of legal AI tools, the ethical rules that apply in the United States and a practical plan to adopt AI responsibly.
Quick answer
Law firms use AI most for legal research, document review and summarization, drafting, contract analysis, e-discovery and practice management tasks such as intake and time entries. The main tools fall into legal-specific platforms (for example Harvey, Thomson Reuters CoCounsel, Lexis+ AI, Clio’s AI features and Spellbook) and general AI assistants used under firm policies. Under ABA Formal Opinion 512 (July 2024), lawyers using generative AI must maintain competence, protect client confidentiality, communicate with clients where appropriate, verify outputs before relying on them, supervise staff and charge reasonable fees. Always verify every citation: courts have sanctioned lawyers for filing AI-invented cases.
This article is general information, not legal or ethics advice. Check the rules and guidance of your own jurisdiction.
AI in the legal profession in numbers
| Source | Date | Key finding |
|---|---|---|
| Clio Legal Trends Report | October 2025 | 79% of legal professionals use AI in some form; adoption is 87% in large firms and 71% among solo practitioners |
| Clio Legal Trends Report | October 2025 | 53% said their firm has no AI policy or they are unaware of one |
| Thomson Reuters, AI in Professional Services Report | February 2026 | Organization-wide generative AI use across professional services reached 40%, up from 22% a year earlier; only 18% of respondents said their organization tracks AI ROI |
| ABA Formal Opinion 512 | July 2024 | First formal ABA ethics opinion on lawyers’ use of generative AI tools |
The gap between widespread individual use and missing firm policies is the main risk area for most firms.
How law firms use AI today
| Task | How AI helps | Human role |
|---|---|---|
| Legal research | Finds relevant cases, statutes and secondary sources from natural language questions and summarizes them | Read the sources and verify every citation |
| Document summarization | Summarizes depositions, contracts, medical records and long files | Check key facts against the originals |
| Drafting | Produces first drafts of letters, motions, memos, clauses and client updates | Edit, apply judgment, ensure accuracy |
| Contract review | Flags non-standard clauses, missing terms and risks against a playbook | Decide on negotiation positions |
| E-discovery | Classifies documents for relevance and privilege at scale | Validate sampling, make privilege calls |
| Due diligence | Extracts key terms across hundreds of agreements | Review exceptions and conclusions |
| Intake and client communication | Answers basic questions, collects information, schedules consultations | Conflicts checks, engagement decisions |
| Practice management | Suggests time entries, drafts invoices narratives, organizes matters | Approve billing and client communications |
Types of legal AI tools
Legal research and drafting platforms
Tools built on legal databases, such as Thomson Reuters CoCounsel (with Westlaw content), Lexis+ AI and vLex Vincent, combine generative AI with curated legal sources, which reduces but does not eliminate the risk of invented authorities. They are usually the safest choice for research because answers link to real sources you can check.
Legal AI assistants for firms
Platforms such as Harvey are built for law firms and legal departments, offering research, drafting and document analysis within security and confidentiality controls designed for legal work.
Contract drafting and review tools
Tools such as Spellbook work inside Microsoft Word to suggest clauses, flag risks and compare contracts against standards. Many contract lifecycle management platforms include similar features for in-house teams.
Practice management with AI
Practice management platforms such as Clio are adding AI features for summarizing matters, drafting documents and communications, and supporting time capture and billing, using data already in the firm’s system.
E-discovery platforms
E-discovery providers use technology-assisted review and generative AI to classify and summarize large document sets.
General AI assistants
General assistants can help with non-confidential tasks such as drafting marketing content, summarizing public information or brainstorming. They should be used only under a firm policy, on business plans with appropriate data protections, and never with client confidential information unless the firm has confirmed the tool’s terms and obtained any required client consent.
We do not rank these tools or accept payment from vendors. Pricing for legal AI is usually per user and often quote based; ask for trials on your own documents.
The ethics rules: what ABA Formal Opinion 512 requires
In July 2024, the American Bar Association issued Formal Opinion 512, its first formal guidance on generative AI. It applies existing Model Rules of Professional Conduct to AI use. Key points:
| Duty | What it means for AI use |
|---|---|
| Competence (Rule 1.1) | Lawyers must understand the capabilities and limitations of the AI tools they use, or work with someone who does, and keep that understanding current |
| Confidentiality (Rule 1.6) | Before entering client information into a tool, lawyers must evaluate the risk of disclosure. For tools that may use inputs to train or share with others, the client’s informed consent is required, and boilerplate consent in engagement letters is not enough |
| Communication (Rule 1.4) | In some situations lawyers should tell clients about AI use, for example when it affects the representation or the client asks |
| Meritorious claims and candor (Rules 3.1, 3.3) | Lawyers must verify AI outputs, including citations and legal analysis, before submitting anything to a court |
| Supervision (Rules 5.1, 5.3) | Firm leaders must set clear policies and train lawyers and staff on AI use |
| Fees (Rule 1.5) | Fees must be reasonable. Lawyers billing by the hour should bill only for time actually spent; firms generally should not charge clients for time spent learning a general tool or for general overhead |
Many state bars have issued their own guidance, and some courts require lawyers to disclose or certify AI use in filings. Check the rules where you practice.
The biggest risks for law firms
1. Hallucinated citations
Generative AI can invent cases, quotes and citations that look real. In Mata v. Avianca (S.D.N.Y. 2023), lawyers were sanctioned for submitting a brief containing fictitious cases generated by a chatbot, and many similar incidents have followed in courts across the country. Rule: never file anything without reading every cited authority in a trusted legal database.
2. Confidentiality and privilege
Entering client information into consumer AI tools may expose it to the vendor or other users, and could raise privilege questions. Rule: use tools with enterprise data protections, no training on your inputs and appropriate security, and get informed client consent where required.
3. Bias and inaccuracy
AI can reflect biases in its training data or miss jurisdiction-specific nuances. Rule: treat AI output as a draft from an inexperienced assistant, not as advice.
4. Billing and client expectations
Clients increasingly ask how AI affects fees, and some issue outside counsel guidelines on AI use. Rule: be transparent, follow client instructions and consider alternative fee arrangements where AI makes work faster.
5. Unauthorized practice and client-facing tools
Chatbots that answer legal questions for the public can create risks if they appear to give legal advice. Rule: design intake tools to collect information and schedule consultations, with clear disclaimers, and route substantive questions to lawyers.
6. Regulatory exposure abroad
Firms with clients or offices in the EU should note that the EU AI Act classifies certain AI used in the administration of justice as high-risk. See EU AI Act explained.
How to adopt AI in a law firm: a step by step plan
- Form a small AI committee with a partner sponsor, IT or security, knowledge management and a risk or ethics lead.
- Write an AI policy covering approved tools, prohibited uses, confidentiality, verification of outputs, client communication and billing. Our guide on how to write an AI acceptable use policy can help.
- Vet vendors for security (for example a SOC 2 report; see what SOC 2 compliance is), data retention, training on customer data, data location and access controls.
- Start with low risk, high value uses: summarizing documents the firm already holds, drafting internal memos, first drafts of routine documents and research with legal-specific tools.
- Train everyone on capabilities, limitations, verification and the policy, and record the training.
- Pilot with a practice group, measure time saved and quality, and collect feedback.
- Update engagement letters and client communication where appropriate, and respect client AI instructions.
- Review quarterly: new guidance, new tools, incidents and results.
Questions to ask legal AI vendors
- Is our data used to train your models or shared with other customers? Can we opt out in the contract?
- Where is data stored and processed, and for how long is it retained?
- What security certifications do you hold (for example a SOC 2 Type 2 report or ISO 27001)?
- How do you prevent and flag invented citations? Do answers link to the underlying sources?
- What content sources does the tool use, and how current are they?
- Can we control access by matter, practice group or ethical wall?
- Is there an audit log of who used the tool and for what?
- How do you handle privileged and confidential documents uploaded by users?
Write the answers into your vendor file. They are part of the reasonable steps expected under your confidentiality duties.
A simple verification checklist for AI-assisted work
- ☐ Every case, statute and quote has been checked in a trusted legal database.
- ☐ Facts have been checked against the source documents.
- ☐ The analysis reflects the correct jurisdiction and current law.
- ☐ No client confidential information was entered into an unapproved tool.
- ☐ A responsible lawyer has reviewed and approved the final work product.
- ☐ Time entries reflect actual time spent.
Measuring the value of AI in a law firm
Useful measures include hours saved on document review and first drafts, turnaround time for client requests, write-offs and write-downs, matter profitability under fixed fees and client satisfaction. Because many firms bill by the hour, efficiency gains can reduce billable hours; firms that benefit most often pair AI with fixed or value-based fees or use saved time for more matters. For a broader method, see how to measure the ROI of AI projects.
Frequently asked questions
Is it ethical for lawyers to use ChatGPT?
It can be, if the lawyer understands the tool, protects client confidentiality, verifies all outputs and complies with ABA Formal Opinion 512 and state rules. Consumer versions may not offer adequate data protections for client information.
Do lawyers need client consent to use AI?
It depends. Under ABA Formal Opinion 512, informed consent is required before entering client information into tools that may disclose it, such as tools that train on inputs. Disclosure may also be appropriate in other situations.
Which AI tool is best for legal research?
Tools built on legal databases, such as CoCounsel, Lexis+ AI or vLex Vincent, are generally more reliable for research because they cite real sources. All outputs still need verification.
Can AI replace paralegals or junior associates?
AI changes their work more than it replaces it, taking over first-pass review and drafting while people focus on judgment, client service and quality control.
How should firms bill for AI-assisted work?
Bill reasonable fees for actual time and value. Do not bill for time saved or for learning general tools; consider fixed fees where AI makes work much faster.
Bottom line
AI can make law firms faster and more responsive, but legal ethics set clear limits. Choose tools with strong confidentiality protections, prefer legal-specific platforms for research, verify every output, write a firm policy, train your people and be transparent with clients about AI and fees. Firms that build these habits now will gain the benefits without the headlines.
Related guides
- EU AI Act explained: what businesses must do
- What is SOC 2 compliance?
- What is RAG? A business guide
- AI for accounting firms: automation and tools
This article is for general information only and is not legal or ethics advice. Consult your state bar’s guidance and qualified ethics counsel.
Sources
- American Bar Association, Formal Opinion 512: Generative Artificial Intelligence Tools (PDF) (July 29, 2024)
- Clio, Legal Trends Report (October 2025), summarized by 2Civility
- Thomson Reuters Institute, 2026 AI in Professional Services Report (February 2026)
- National Conference of Bar Examiners, ABA Formal Opinion 512 provides needed guidance on lawyers’ use of GAI (2024)