Thomson Reuters (Westlaw's owner) and LexisNexis have marketed their legal-specific AI tools—Westlaw AI-Assisted Research, Lexis+ AI, and Ask Practical Law AI—as safer alternatives to general-purpose models. Stanford research reveals they still hallucinate between 17% and 34% of the time, compared to 58% to 88% for general AI tools on legal queries. The "WL" citation problem stems from these tools mixing real and fabricated elements to create plausible-looking but factually false citations. Hallucinated citations are now appearing in databases like Google Scholar, making detection harder.
The timing matters. Courts and bar associations are enforcing stricter penalties for hallucinated content, including potential attorney suspensions. The ABA's Formal Opinion 512 explicitly holds lawyers fully responsible for verifying all AI-generated work products and citations—making human oversight an ethical duty, not optional. Without systemic changes and rigorous verification protocols, the legal profession faces a flood of fabricated authority that could compromise judicial decisions and expose practitioners to professional sanctions.