OPERATIONS
AI tools for lawyers: what they do well and where to be careful
A practical overview of AI for law firms, from general assistants and legal research tools to free options, with the confidentiality and accuracy rules lawyers need to follow.
By Personal Injury Leads Editorial Team3 min read
AI tools are now part of ordinary legal work. Used well, they save hours on drafting, summarising and organising. Used carelessly, they have led to sanctions. This guide sets out the main types of tool, what each is good for, and the professional duties that apply.
The main types of AI tool
| Type | Examples | Best used for |
|---|---|---|
| General assistants | Claude, ChatGPT, Gemini, Microsoft Copilot | Drafting, summarising, restructuring, explaining, brainstorming |
| Legal research AI | Tools built into legal research platforms such as Westlaw and Lexis | Finding and summarising authority linked to a verified database |
| Document review and drafting | Contract review, discovery review and document automation tools | Classifying and extracting from large sets of documents |
| Transcription and notes | Meeting and deposition summarisers | Turning recordings into searchable text and summaries |
| Practice software features | AI built into case management systems | Summarising records and drafting from case data |
Products and features change quickly. Treat the examples as categories to explore, not recommendations.
What general assistants are good at
Tools such as Claude and ChatGPT are useful wherever a lawyer already knows the answer or can check it.
- Turning rough notes into a first draft of a letter or memo
- Summarising a long document you will then read yourself
- Rewriting a paragraph in plainer language for a client
- Building a chronology from documents you supply
- Preparing questions for a deposition or client meeting
- Explaining an unfamiliar industry or technical subject
They are weaker where accuracy cannot be checked easily. A general assistant can state a rule confidently and be wrong, and can produce a citation to a case that does not exist.
The accuracy problem
In 2023 a federal court in New York sanctioned lawyers in Mata v. Avianca for filing a brief that cited cases invented by an AI tool. Similar incidents have followed in other courts.
The lesson is narrow and firm: verify every citation and every statement of law against a primary source before it leaves the office. A tool connected to a real legal database reduces the risk. It does not remove the duty to check.
Some courts and individual judges now require lawyers to disclose or certify their use of AI in filings. Check the rules and standing orders of the court you are in.
Confidentiality comes first
Before entering anything about a client into an AI tool, find out what the provider does with it.
- Is your input used to train the provider's models?
- Who at the provider can see it?
- How long is it kept, and can you delete it?
- Is there a business or enterprise plan with stronger protections?
The American Bar Association addressed this in Formal Opinion 512 in 2024. In summary, lawyers using generative AI must understand the tool well enough to use it competently, protect client confidences, supervise its use, communicate with clients where appropriate, and charge reasonable fees for work done with it. Many state bars have issued their own guidance.
Are free legal AI tools worth using?
Free versions of general assistants are capable enough for non-confidential tasks: drafting a blog post, tidying an internal policy, outlining a presentation.
For client work, the free tier is usually the wrong choice. Consumer plans often allow the provider to use your conversations to improve its models unless you opt out, and they lack the administrative controls a firm needs. Paid business plans generally offer stronger terms. Read them before relying on them.
A simple policy for a small firm
- Approve specific tools. List what staff may use and for what.
- No client-identifying information in any tool the firm has not approved for confidential work.
- Everything is a draft. A lawyer reviews and takes responsibility for anything an AI tool produced.
- Check every citation against the source.
- Follow the court's rules on disclosure.
- Review the policy every few months, because the tools change.
Where AI fits in intake, and where it does not
AI is useful around intake: transcribing calls, summarising notes, checking that a file is complete. It is less suited to the call itself. A person who has just been injured wants to talk to someone who listens, asks sensible follow-up questions and can do so in their own language.
That is why our screening is done by people. Our in-house intake specialists answer in English and Spanish and decide whether a call qualifies before it becomes a live transfer or a signed retainer.
This article is general information for law firms, not legal advice. Rules on attorney advertising and practice differ by state; check your own jurisdiction before acting on it.