In the news
Cited but Not Consulted: A Counterfactual Audit of Legal Chain-of-Thought Faithfulness
arXiv cs.AI · Published · 3 min read
In 30 seconds
- What happened
- Researchers found that legal AI models cite correct statutes but don't actually use them to reach verdicts, showing poor faithfulness in legal reasoning.
- Why it matters
- Matters for engineers building legal AI systems and organizations relying on model explanations for compliance, audit, or decision justification.
- Watch out
- The gap persists across model sizes and specialized legal models, and verdicts remain vulnerable to adversarial manipulation hidden in case facts.
- language model
- reasoning
- benchmark
- open-weight
The patterns behind this
Each one covers how the technique works, when it earns its cost, and where it breaks.
The Agent Architect
One pattern, one tradeoff, one production failure story. A short weekly briefing for people building agentic systems.
Weekly email, one-click unsubscribe. We only use your address to send the briefing.