AI for Lawyers: The Legal AI Tools Worth Using in 2026

Last updated: 26 October 2026. This guide draws on ABA Formal Opinion 512, the Solicitors Regulation Authority’s Risk Outlook report on AI, the Law Society’s generative AI guidance, the Mata v. Avianca and Ayinde v Haringey judgments, Stanford research on legal AI accuracy and official vendor pages. It is general information, not legal advice.

Short answer: AI for lawyers now falls into five useful groups: research assistants built on legal databases, drafting and contract review tools, eDiscovery review, AI inside practice management software, and general chatbots sold on enterprise terms. All of them can be wrong. US and English rules let lawyers use them, provided client data is protected and every output is checked.

Key takeaways

• The safest legal AI tools are grounded in a trusted source, such as Westlaw, Lexis or your own documents, and show citations you can check.
• ABA Formal Opinion 512 (29 July 2024) applies the existing duties of competence, confidentiality, communication, supervision and reasonable fees to generative AI. It does not ban any tool.
• In England and Wales, the SRA says firms stay responsible for AI output, and the High Court has warned that free tools such as ChatGPT cannot do reliable legal research.
• Fake citations still lead to sanctions. Pilot any tool on real but low-risk work, measure the results, and write the rules down before a wider rollout.

What can AI do for lawyers today?

Most of the value sits in tasks that involve reading or producing large amounts of text. Current tools can summarise cases and documents, answer research questions with linked authority, produce first drafts of letters, clauses and motions, compare a contract against a playbook, sort millions of documents for relevance in disclosure, and pull dates and tasks out of correspondence into a matter file.

What they cannot do is take responsibility. A generative AI model predicts likely text, so it can produce a confident answer that is wrong, misread a document or invent an authority. The professional rules described below all start from that point: the lawyer remains accountable for anything that goes to a client or a court. For a running overview of products, see our legal AI software directory.

Which legal AI tools are worth knowing?

The market changes monthly, so it helps to think in categories first and brands second. The examples below are named for what their makers say they do, not as endorsements.

Research assistants matter most for litigators, because they are tied to citators and primary law. Our CoCounsel review and our comparison of AI research platforms go into detail. Transactional teams usually get faster returns from review tools that work against a clause playbook; our guide to AI contract review software explains how those products differ. Thomson Reuters describes CoCounsel Legal’s agentic features in its launch announcement.

Can lawyers use ChatGPT and other general AI tools?

Yes, but the version and the task matter. Free consumer chatbots are a poor fit for client work for two reasons. First, the terms may allow the provider to use what you type, and the Law Society’s guidance tells solicitors not to feed confidential information into generative AI tools they do not control. Second, a general model is not connected to a citator, so it has no reliable way to know whether a case exists or is still good law.

Enterprise plans change the confidentiality picture, not the accuracy picture. OpenAI’s business data page says it does not train on organisations’ data by default, lets qualifying customers set retention periods and offers data residency to eligible Enterprise customers. That makes ChatGPT for lawyers a reasonable option for summarising a document you already hold, rewriting a client email in plain English or brainstorming cross-examination themes. It is still the wrong tool for finding authority. Treat a general AI legal assistant like a fast, capable trainee with no access to a law library.

What do US ethics rules say about AI?

The main source is ABA Formal Opinion 512, issued on 29 July 2024. The ABA’s announcement explains that it applies the Model Rules on competence (1.1), confidentiality (1.6), communication (1.4) and fees (1.5) to generative AI, and the opinion also addresses supervisory duties. The key points:

• Competence: lawyers need a reasonable understanding of a tool’s capabilities and limits, and the level of checking depends on the tool and the task.
• Confidentiality: before putting client information into a self-learning tool, a lawyer needs the client’s informed consent, and general boiler-plate wording in an engagement letter is not enough.
• Fees: a lawyer may bill for time spent entering information and reviewing the output, but in most circumstances may not charge a client for learning how to use the tool.
• Supervision: managing lawyers should set clear firm policies on permitted AI use.

State bars have issued their own opinions, so check your jurisdiction. Some judges also require disclosure. In 2023 Judge Brantley Starr of the Northern District of Texas began requiring lawyers to certify either that no generative AI was used in a filing or that any AI-drafted language was checked by a human, and said filings without the certificate would be struck. Our explainer on ABA Formal Opinion 512 covers the opinion in full.

What are the rules in England and Wales?

There is no AI-specific rulebook for solicitors. The SRA Principles and Codes of Conduct apply as they do to any other work. The SRA’s Risk Outlook report on AI, published on 20 November 2023, says firms remain responsible for AI output even when a third party supplies the tool, should supervise both the systems and staff use of them, and should “not trust an AI system to judge its own accuracy”.

The Law Society’s Generative AI: the essentials, updated in September 2025, adds practical advice: fact-check outputs, verify every citation against an authoritative source, use fictional data when testing, and remember that misleading a client or the court breaches paragraph 1.4 of the SRA Code of Conduct. It also notes that the SRA has no specific guidance on disclosing AI use to clients.

The regulator is not hostile to the technology. In May 2025 the SRA authorised Garfield.Law, an AI-driven firm that helps small businesses recover debts in small claims of up to £10,000, after checking its quality controls. The firm’s system does not propose case law, and named solicitors remain accountable. The judiciary also publishes its own AI guidance for judges, most recently refreshed in October 2025.

What happens when AI gets it wrong?

The best-known US case is Mata v. Avianca. On 22 June 2023 Judge P. Kevin Castel of the Southern District of New York fined two lawyers and their firm $5,000 after a brief cited six non-existent opinions generated by ChatGPT, and ordered them to write to each judge falsely named as an author, as the ABA Journal reported. The judge stressed that the lawyers “doubled down” rather than correcting the record.

In England, the Divisional Court’s judgment in Ayinde v London Borough of Haringey [2025] EWHC 1383 (Admin), handed down on 6 June 2025, said that freely available generative AI tools “are not capable of conducting reliable legal research”. In Ayinde, grounds for judicial review cited five cases that did not exist; the barrister and the law centre were each ordered to pay £2,000 in wasted costs and referred to their regulators. The court said that placing false citations before it is likely to make a referral to the regulator appropriate.

Specialist tools reduce the risk but do not remove it. A Stanford study published in May 2024 found that legal research tools hallucinated on more than 17% of benchmark queries for Lexis+ AI and Ask Practical Law AI, and more than 34% for Westlaw AI-Assisted Research. The products have changed since, but the lesson holds. Our analysis of AI hallucination sanctions tracks how courts are responding.

How should a firm choose and pilot a legal AI tool?

Start with a problem, not a product. The firms that get value pick two or three high-volume tasks, such as first-pass NDA review or deposition summaries, and test tools against them.

1. Define the task and the baseline. Record how long the work takes now and what a good result looks like.
2. Check the data terms. Ask where data is stored, how long it is kept, whether it trains any model and who at the vendor can see it. Get answers in the contract, not the sales deck.
3. Prefer grounded tools. For research, choose a tool connected to primary law and a citator. For contracts, choose one that works against your own templates and playbooks.
4. Run a time-limited pilot. Use a small group, real but low-risk matters and a scoring sheet for accuracy, time saved and errors caught.
5. Write the rules down. Approved tools, banned uses, verification steps, client consent and billing. Our law firm AI policy template gives a structure.
6. Decide, then train. Roll out, renegotiate or stop. Our guide to running a legal AI pilot covers how to reach a clear decision.

Price varies widely, from per-user subscriptions to enterprise contracts, so include training time and integration costs in any comparison.

Frequently asked questions

What is the best AI for lawyers?

There is no single best tool. For research, tools tied to Westlaw or Lexis are the most defensible because they link to a citator. For contracts, review tools that apply your playbook tend to deliver faster gains. General chatbots on enterprise terms suit drafting help and summaries.

Is it ethical for lawyers to use ChatGPT?

It can be. ABA Formal Opinion 512 and SRA guidance allow AI use if lawyers stay competent with the tool, protect client confidences, check the output and bill fairly. Putting client data into a free consumer version without consent is the main risk.

Do lawyers have to tell clients they use AI?

Sometimes. Under Opinion 512, informed consent is needed before client information goes into a self-learning tool, and boiler-plate engagement letter wording is not enough. In England and Wales, the SRA has no specific rule, but its 2023 report says firms should tell clients when AI is used on their case.

Do courts require lawyers to disclose AI use?

Some do. Individual US judges, such as Judge Brantley Starr in Texas, have issued standing orders requiring certification about AI use. Always check the local rules and the assigned judge’s standing orders before filing.

Can AI legal research tools still invent cases?

Yes. A 2024 Stanford study found errors on more than one in six benchmark queries even in leading legal research products. Every citation should be opened, read and checked in a citator before it goes into a filing.

Can lawyers bill for time saved by AI?

Not on an hourly basis. Opinion 512 says lawyers may bill for time spent prompting and reviewing AI output, but not for hours the work would have taken without it, and usually not for learning how to use the tool.

This article is general information about legal technology and professional conduct rules and is not legal advice. Ethics rules and court orders differ between jurisdictions and change often, so check your own bar, regulator and court rules before relying on any AI tool.