LePhamtomCite: Legal Citation Hallucination Benchmark

LePhamtomCite is a benchmarking dataset for evaluating AI systems on legal citation hallucination detection. It was created by the ai-law-society-lab and last updated on April 27, 2026. The dataset addresses the problem of fabricated or misrepresented case citations in court filings.

Use Cases

• Benchmarking AI systems for detecting fabricated legal citations based on the dataset's structured evaluation framework.
• Training models to identify citation hallucinations in AI-generated legal filings.
• Evaluating the reliability of large language models in legal drafting tasks.

Strengths

• Dataset is explicitly designed as a structured benchmark for a specific, high-stakes problem in legal AI.
• Authored by a named research lab (ai-law-society-lab), suggesting an academic or research-oriented purpose.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for large-scale training.

Provenance

Source

ai-law-society-lab

Freshness

Last updated 2026-04-27 15:11:46; freshness should be verified.