Can an AI chatbot give legal advice?
By Moshe Dor, FounderLast updated
Where is the line between information and advice?
Generality. “Florida’s personal injury statute of limitations is generally two years” is information. “You still have time to sue for your accident” is advice — it applies law to the visitor’s facts. Intake bots must live entirely on the information side and be tested against attempts to drag them across.
How is a bot fenced against giving advice?
Defense in depth: a system prompt that defines refusal behaviors, response templates that redirect facts-specific questions to consultation booking, adversarial pre-launch testing (“just tell me if I have a case”), and conversation-log review. FirmCited ships intake bots with these fences as the default, not an option.
What happens when a visitor pushes for advice?
The bot acknowledges the question matters, states plainly that it cannot advise because it is not a lawyer, and converts the moment: “That is exactly what a consultation answers — I can get your details to the team now, and you will hear back in under a minute.” The refusal becomes the conversion.
| Visitor asks | Fenced bot response |
|---|---|
| What does this firm charge? | Answers directly — administrative |
| What is a contingency fee? | Explains generally — educational |
| Do I have a case? | Declines + offers consultation — advice line |
| Should I sign this settlement? | Declines + urgent human escalation |
See where your firm stands.
The $490 AI Visibility Audit answers this question about your firm, with verbatim engine results — 25 prompts, 4 engines, under 48 hours. Credits toward any install.