Analysis
A small online tool called CiteProof is positioning itself to catch fabricated case citations and misquoted court opinions in legal filings, at a moment when AI-generated hallucinations keep showing up in real court documents, according to The Register. The company describes itself as a small operation built with some "vibe-coding," and the report names no founder.
The tool's core design choice is deliberately anti-AI: CiteProof says it uses no LLMs or probabilistic models to do the checking, instead running deterministic searches against public court records and sending only citation strings -- never full case text -- to CourtListener for verification. "You cannot reliably catch AI hallucinations with more AI guesswork," a company spokesperson said. The tool checks whether a citation exists and whether a quote matches the source; it does not assess the quality of the underlying legal argument.
“A Georgia court order was vacated in 2025 after it was found to cite fictional cases.”
The problem it addresses is well documented. A Georgia court order was vacated in 2025 after it was found to cite fictional cases. Even Anthropic's own outside lawyers were caught in a music-publishing lawsuit submitting a declaration containing Claude-formatted citations that pointed to improper records -- an error the lawyers themselves missed in review before filing. GPT-4 passed a bar exam in 2023 but still hallucinated case law in practical use, underscoring that passing a test and producing reliable citations in live filings are different problems entirely.
Pricing is modest and firm-oriented: free users can check individual citations or statutes, a single brief scan costs $49, a five-pack runs $199, and law firms can pay $499 a month for 200 brief reviews with full verification reports, quotation checking and bulk scanning. CiteProof says the only data leaving its servers is citation strings sent to CourtListener, with everything else processed in the US and nothing routed through an LLM -- a privacy and liability pitch aimed squarely at risk-averse law firms rather than individual lawyers.
The company itself is candid that AI adoption in legal drafting isn't slowing down regardless of the hallucination risk: "lawyers and associates are already using AI to speed up research and drafting, and that adoption isn't stopping," its spokesperson said. That's the real market CiteProof and any competitor is chasing -- not replacing AI drafting tools, but selling the verification layer firms will increasingly be expected to run before anything gets filed.
