Legal research is where AI's fabricated-authority risk is sharpest, and where the right tool makes the biggest difference. The distinction that matters is what the tool is built on.
General models against legal tools
A general chatbot holds no reliable index of cases or statutes, so it can invent them. A research tool built on a known legal database works from real sources and links to them, so you can open the authority and read it. The first belongs nowhere near legal research. The second can save real time.
Verify, always
Even with a proper tool, treat the summary as a pointer rather than the answer. Open the case, read the passage, and check that it still represents the law. A summary that is ninety per cent right is still wrong where it counts.
Keep the trail
Note the sources you relied on and the date you checked them. If a position is later questioned, you can show your working. The cost of skipping this step is set out in stopping AI mistakes reaching your advice.
AI can make research faster. It cannot decide what the law is, and it cannot answer for getting it wrong. That stays with you.
The judiciary’s AI guidance warns about fabricated authorities in much the same terms.
If your researchers use a general chatbot because nobody has chosen the proper tool, that choice is overdue, and we help firms make it well: get in touch.