The comparison
ChatGPT makes up cases.
CommonBench cites real ones.
Generic AI chatbots hallucinate case law. Lawyers have been sanctioned for relying on them. CommonBench is built differently — citations are checked for existence against a 66,200+ authority corpus (39,000+ verified for retrieval) and, where source policy permits, live public endpoints; unconfirmed citations are flagged, and failures are stripped. Status is always shown because no automated check is infallible.
Feature comparison: generic AI vs CommonBench
| Feature | ChatGPT / Generic AI | CommonBench |
|---|---|---|
| Case citations | Frequently hallucinated | Checked for existence; verified / flagged / stripped |
| Jurisdiction awareness | Generic, US-biased | UK, HK, SG, AU, US — correct terminology and procedure |
| Procedural pathway | Doesn't tell you what to file | Procedural orientation — court, form, deadline; verify specifics with the court |
| Opponent analysis | Not available | Defence anticipation + weakness mapping |
| Case strength score | Not available | Indicative 0–100 score with weighted factors — a research aid, not counsel's judgment |
| Costs estimate | Vague or absent | Broken down by litigation stage + costs exposure |
| Limitation calculator | Not available | Specific dates and deadline warnings |
| Free ADR alternatives | Not flagged | Ombudsmen, tribunals, and free schemes identified |
| Court sanctions risk | Multiple lawyers sanctioned for fake citations | Citations checked for existence; status shown before you rely |
On the record
When AI citations go wrong
Real cases where lawyers relied on AI-generated legal research with fabricated citations.
Mata v Avianca (2023) — US Federal Court
“Technological advances are commonplace, and there is nothing inherently improper about using a reliable artificial intelligence tool for assistance. But existing rules impose a gatekeeping role on attorneys to ensure the accuracy of their filings.”
— Judge P. Kevin Castel, Mata v. Avianca, Inc., 678 F.Supp.3d 443 (S.D.N.Y. 2023)
Park v Kim (2024) — US Second Circuit
The Second Circuit referred an attorney for disciplinary proceedings after she submitted a brief citing a nonexistent case generated by ChatGPT. The court held that the use of AI tools “does not excuse an attorney from separately ensuring that submissions to the Court are accurate or legally tenable.”
Park v. Kim, No. 22-2057 (2d Cir. 2024)
UK Solicitors Regulation Authority
The SRA advises that solicitors using AI tools remain fully bound by the SRA Code of Conduct, including obligations of competence and service. Solicitors must review and verify AI-generated outputs before relying on them or sharing them with clients or courts.
SRA, Compliance tips for solicitors regarding the use of AI and technology (Solicitors Regulation Authority)
CommonBench is engineered against exactly this failure: citations are checked for existence against the corpus (and permitted live sources) before display, unconfirmed citations are flagged, and failures are stripped. Existence checks do not prove good law or proposition accuracy — which is why the verification status is always shown, so you can confirm before you rely.
Legal research you can verify.
Same question, five jurisdictions, citations checked for existence before they reach you.