Hallucination Risk and Escalation
When to trust AI and when to escalate to a human.
Hallucination Risk and Escalation
AI can invent facts, citations, and legal conclusions. Know when to trust it and when to escalate.
What is hallucination?
Models sometimes generate plausible-sounding but false information. They may invent case names, statute sections, or dates. They do not "know" the law—they predict text.
When to verify everything
- Legal conclusions: Never rely on AI for final legal advice.
- Citations: Always look up statutes and cases yourself.
- Deadlines and procedures: Confirm against court rules and local practice.
- Client-specific facts: AI has no access to your matter. Verify all matter details.
When to escalate
- Unclear or conflicting outputs
- High-stakes decisions (litigation, transactions, regulatory)
- Jurisdiction-specific questions
- Anything you would normally run by a senior attorney
Use AI to speed up research and drafting. Use humans to validate and decide.