Asking a Chatbot About Your Injury Case? What AI Gets Wrong About DC, Maryland, and Virginia Law
Let's be honest: if you were hurt in a crash last week, there is a decent chance you have already asked an AI chatbot about it. Millions of people now type questions like "how much is my car accident case worth" or "can I sue if I was partly at fault" into an AI assistant before they ever talk to a lawyer. We are not here to scold anyone for that. The technology is genuinely useful for understanding legal vocabulary, organizing questions, and getting oriented.
But we have now read enough AI-generated "legal advice" — some of it brought to us by prospective clients — to notice a pattern. Chatbots are trained mostly on the law of everywhere, which means they default to the law of nowhere in particular. And the DC, Maryland, and Virginia region is precisely where the law of nowhere in particular will hurt you, because this region's injury law is nationally unusual. Here are the mistakes we see most, without naming any particular product — they all make versions of these errors.
Mistake 1: "Your compensation will just be reduced by your percentage of fault."
This is the big one. Most states use comparative negligence, so most AI answers confidently explain that being 20% at fault means recovering 80% of your damages. DC, Maryland, and Virginia are three of the tiny handful of American jurisdictions that still follow pure contributory negligence — where being even 1% at fault can mean recovering nothing. Maryland's highest court reaffirmed the rule in Coleman v. Soccer Association of Columbia in 2013, and it remains the law in all three jurisdictions today. An injured person who accepts the chatbot's framing may casually admit "partial fault" to an adjuster — and hand the insurer a complete defense.
Mistake 2: Missing the local exceptions that save cases.
The flip side is just as damaging. Generic answers rarely know that DC abolished the contributory negligence bar for pedestrians and cyclists struck by cars (the Motor Vehicle Collision Recovery Act of 2016), and extended that protection to e-scooter and motorcycle riders as "vulnerable users" in 2021. They rarely mention the last clear chance doctrine, which can rescue a claim even where the injured person was negligent. A chatbot that tells a DC cyclist "you were partly at fault, you have no case" is wrong twice.
Mistake 3: Silence about the notice traps.
Ask an AI about deadlines and you will hear about the statute of limitations — usually the right number, three years in DC and Maryland, two in Virginia for most injury claims. What the generic answer misses are the short-fuse notice requirements that quietly kill cases against governments:
| Trap | The rule |
|---|---|
| Suing the District of Columbia | Written notice to the Mayor within 6 months (D.C. Code § 12-309) — one of the harshest notice rules in the country |
| Suing a Maryland county, city, or local agency | Notice within 1 year under the Local Government Tort Claims Act |
| Suing Metro (WMATA), a state agency, a school board | Each carries its own immunity and procedural wrinkles that generic answers gloss over |
A claim can be fully alive under the statute of limitations and already dead for lack of notice. We have never seen a chatbot volunteer § 12-309 unprompted.
Mistake 4: Out-of-date law, delivered confidently.
AI models have training cutoffs, and the law keeps moving. Recent examples from our own region: Virginia abolished its underinsured-motorist "offset" for policies issued or renewed since July 2023 — a change that can be worth six figures to a seriously injured crash victim, and one that outdated answers still get wrong. Maryland renamed its appellate courts in December 2022, so citations to the "Court of Appeals of Maryland" now confuse both humans and machines. Damages caps change annually in Virginia med mal cases and have changed twice in two years for Maryland abuse claims. A chatbot rarely tells you when its knowledge ends.
Mistake 5: Invented specifics.
Ask for case citations or settlement values and generative AI will sometimes simply make them up — fluently, with realistic-looking case names and dollar figures. Courts around the country have sanctioned lawyers for filing AI-invented citations. If a machine will hallucinate to a judge's face, it will certainly hallucinate to yours.
What this means for injured people in the DMV
Use AI the way you would use a knowledgeable friend who has never set foot in a DMV courtroom: good for vocabulary, background, and questions to ask — bad for decisions. Specifically:
- Never admit or apportion fault based on a chatbot's comparative-negligence framing. In this region, fault talk is case-ending talk.
- Calendar nothing based on an AI answer. Deadlines here include six-month and one-year traps the generic answer omits.
- Verify everything jurisdiction-specific. If the answer doesn't distinguish DC from Maryland from Virginia, it isn't an answer about your case.
- The stakes are asymmetric. A free consultation with a human lawyer costs you nothing. A confident wrong answer can cost you the case.
We are happy to be your second opinion on anything a chatbot told you — that conversation is free, and unlike the chatbot, we are accountable for what we tell you.
If you or a loved one has been hurt, call Posey Lebowitz at (202) 524-0123 or send us a message for a free consultation.
Part of our DMV case law archive, compiled in 2026 to reflect the archive of some of the most important decisions in injury law.
Sources
- Coleman v. Soccer Association of Columbia, 432 Md. 679 (2013) (CourtListener)
- D.C. Code § 50-2204.52 (Motor Vehicle Collision Recovery Act comparative standard)
- D.C. Law 23-183 (Vulnerable User Collision Recovery Amendment Act)
- D.C. Code § 12-309 (notice to the District)
- Va. Code § 38.2-2206 (UM/UIM coverage, as amended by SB 754 (2022))