Law is a profession built on reading, writing and reasoning about text — exactly what language models do well. Law firms and legal departments have adopted AI quickly for research, document review and drafting. The profession also produced one of the best-known cautionary tales about AI.

Where AI is used

AI tools search case law, statutes and regulations and summarize relevant authorities. Specialized legal research platforms combine language models with curated legal databases and citations.

Contract review and drafting

AI extracts key clauses, flags unusual terms, compares contracts against playbooks and drafts first versions. In-house legal teams use it to handle high volumes of routine agreements such as non-disclosure agreements.

E-discovery and document review

Litigation can involve millions of documents. Machine learning has been used for years in “technology-assisted review” to identify relevant documents; generative AI adds summarization and question answering over document sets.

Due diligence

In mergers and acquisitions, AI reviews data rooms to find risks across many contracts.

Client communication and knowledge management

Drafting emails, summarizing matters, and searching a firm’s own precedents and know-how.

Access to justice

AI tools can help people understand legal problems, forms and procedures — potentially widening access for those who cannot afford lawyers, with risks if advice is wrong.

The hallucination problem

In 2023, in the US case Mata v. Avianca, lawyers submitted a brief citing court decisions that did not exist — fabricated by ChatGPT. The court sanctioned the lawyers. Since then, courts in many countries have encountered filings with invented citations or misquoted authorities, leading to sanctions and new court rules requiring disclosure or verification of AI use.

The lesson: general-purpose chatbots can produce plausible but fictional case law. Every citation must be checked against authoritative sources.

Ethics rules

Professional duties apply regardless of tools. In July 2024, the American Bar Association issued Formal Opinion 512, its first ethics guidance on generative AI, addressing:

  • competence: understanding the benefits and risks of the tools used;
  • confidentiality: protecting client information entered into AI tools;
  • communication: informing clients about AI use where appropriate;
  • supervision: overseeing AI use by staff;
  • fees: billing fairly when AI saves time.

Bar associations and courts in other countries have issued similar guidance.

What changes for the profession

Routine drafting and review become faster, putting pressure on billing by the hour and on junior lawyer training, which traditionally relied on exactly those tasks. Value shifts toward judgment, strategy, negotiation and client relationships.

Courts have been clear that responsibility for filings stays with the lawyer. A basic verification routine:

  1. Check every citation exists in an authoritative database, not by asking the AI again.
  2. Read the actual decision to confirm it says what the AI claims.
  3. Check it is still good law — not overturned, superseded or limited.
  4. Confirm quotations word for word against the source.
  5. Check jurisdiction and dates to ensure the authority applies.
  6. Record the tools used in case a court rule requires disclosure.

Legal research platforms that ground answers in curated databases reduce, but do not eliminate, these risks; independent studies have found that even specialized legal AI tools can produce incorrect or misleading answers.

What to watch next

  • Court rules on disclosure and certification of AI use in filings.
  • Sanctions cases involving AI-fabricated citations.
  • Changes in law firm billing models as AI reduces hours.
  • Regulation of AI legal advice tools for consumers.

Key takeaways

  • Lawyers use AI for research, contracts, e-discovery, due diligence and drafting.
  • In Mata v. Avianca, lawyers were sanctioned for submitting fake ChatGPT-generated citations.
  • Every AI-provided citation must be verified against authoritative sources.
  • ABA Formal Opinion 512 (July 2024) addresses competence, confidentiality, communication, supervision and fees.
  • AI pressures hourly billing and changes how junior lawyers learn.

Frequently asked questions

It can help you understand concepts and prepare questions, but it can be wrong and doesn’t replace a qualified lawyer, especially for important matters.

Do lawyers have to tell clients they use AI?

It depends on the jurisdiction and circumstances. Ethics guidance generally calls for communication when AI use affects the representation or fees.

Why do AI tools invent cases?

Language models generate text that resembles legal citations based on patterns, without checking a database, unless they are connected to verified legal sources.