Who the AI names when families ask about medical malpractice
We asked Google and ChatGPT the questions patients and families ask after something goes wrong in their care, in ten US cities. Here the two assistants named the same firms for more questions than in any other practice we've studied, and both often picked specialists with a page for that exact injury.
What we found
- Google's AI named firms for 8 of 11 families' questions, and ChatGPT for 10 of its 12, which included a second Denver question.
- The two assistants named the same firms for more questions here than in any other practice we've studied. In Phoenix, all 3 of ChatGPT's firms were on Google's list; in Seattle, 4 of its 5; in Boise, 3 of its 4.
- The named firms usually had a page for that exact harm: a medication error, a brain injury at birth, bedsores in a nursing home, a missed diagnosis.
- The AI list and the map showed different firms. On the Seattle search, the med-mal firms in Google's AI list had between 7 and 58 reviews, and the map below it showed injury firms with about 700 to 2,000.
The question
When a patient or a family asks Google or ChatGPT about a medical mistake, and who to call about it, which firms get named? And what do the named firms have that others don't?
Med-mal firms often get their cases from other lawyers rather than from advertising. A family that asks an AI first is a door a referral-driven firm may never have checked.
What Google's AI did
Google gave an AI answer to all 11 family questions and named firms in 8 of them. It named none for the New York stroke, the Atlanta missed cancer, or the Denver family turned down three times. In New York and Atlanta it still drew on firms' pages about that exact harm, a missed stroke in the ER and breast-cancer misdiagnosis, without listing the firms as ones to call.
Where it named firms in answer to a family's situation, the firms usually had a page for that exact harm: medication-error and hospital-error pages in Charlotte, pages about brain injury at birth in Chicago, bedsore and nursing-home pages in Phoenix, misdiagnosis pages in Boise. National brands showed up mostly as sources. Google named mostly local firms, specialists among them, though in Phoenix and Seattle it also named big injury firms.
The answers told families what to look for, and it reads like a checklist for a firm's own website:
- A specialist. "You should strictly consult a specialized medical malpractice attorney in the Los Angeles area rather than a general personal injury lawyer."
- Who reads the records. "Top firms routinely work with independent medical experts and employ nurses or doctors on staff to review charts before taking a case."
- Who does the work. Among the questions it gave the Chicago parents to ask: "Who will actually be handling day-to-day work on our case?"
For the Denver family turned down three times, Google said lawyers turn down the vast majority of cases, explained the cost of bringing a case, and told the family to ask each firm for its specific reason. Among the pages it drew on were firms' own pages about why cases get turned down.
What ChatGPT did
ChatGPT named firms for 10 of the 12 family questions, counting the second Denver question. It named none for the Atlanta missed cancer or the Kansas City cost question, though it used one Kansas City firm's page about how long cases take as a source.
Where both named firms, they mostly agreed:
| Market | Firms both assistants named |
|---|---|
| Phoenix, bedsores | 3 of ChatGPT's 3 |
| Seattle, best med-mal lawyer | 4 of ChatGPT's 5 |
| Boise, missed sepsis | 3 of ChatGPT's 4 |
| Charlotte, medication error | 2 of ChatGPT's 3 |
| Los Angeles, nerve damage | 1 |
| Los Angeles, in Spanish | 1 |
| Chicago, brain injury at birth | none of ChatGPT's 4 |
No other practice we've studied had as many questions where both assistants named the same firms. In Chicago, ChatGPT's four were different firms, though two of them had pages about that injury near the top of Google's results.
It chose firms with a page for that injury, and said so. One firm "has a dedicated bedsore practice page," it said in Phoenix, and in Charlotte, "Its website has a specific page about medication-error claims."
It left out the high-review injury firms. Google's Phoenix table included a large personal-injury firm with more than 2,500 reviews. ChatGPT named the other three and left that one out. On the Seattle search it dropped the two injury firms with 733 and about 2,000 reviews, and called the malpractice-only firm "my first call."
It named Denver firms where Google named none. Four for the first Denver question and five for the bowel injury, three of them the same both times. Gemini, given the bowel-injury question, named three Denver firms, and none of them were among ChatGPT's five.
Its advice matched Google's. See a specialist "rather than a general personal injury lawyer," ask who reviews the records, ask who advances the costs, and ask who will handle the case.
The map and Google's AI list
The Seattle search shows a med-mal specialist's problem in one picture. Google's AI answer listed six firms. The three firms on the map below it were injury firms with about 700 to 2,000 Google reviews. The med-mal firms in the AI's list had between 7 and 58.
A med-mal boutique sees far fewer clients than an injury firm, and its best results are often confidential. Google's AI answer put the specialists in front of the person searching. The map didn't.
Spanish
Asked in Spanish for a Los Angeles malpractice lawyer who speaks Spanish, for a baby with cerebral palsy, Google answered in Spanish and named four firms. All four were named for their Spanish service, and each had a Spanish page for that injury. One was described as offering a free review of the records in Spanish.
ChatGPT answered in Spanish too and named four firms, each for its Spanish service and its birth-injury page. One was the same as Google's. Every firm either assistant named had a Spanish page for that exact injury.
How the assistants answer the firm's own question
We also asked how medical-malpractice firms get more cases. Google's AI said firms get cases "primarily through attorney referrals, targeted digital marketing, and past client networks," drawing on a lawyers' forum thread and a trial lawyers' magazine article, among other sources.
ChatGPT answered from its own knowledge, with no search: build referrals from other lawyers and local search, test paid search, and judge each channel by what a signed case costs. It never mentioned being named in AI answers.
What it means for a med-mal firm owner
A page for each injury you take, in your state. A missed diagnosis, a surgical injury, a medication error, a birth injury, a bedsore. A page for that exact harm is what most of the named firms had.
Say who reviews the records. Both assistants told families to ask about nurses, doctors, and medical experts. If your firm has them, your site should say so plainly, along with who will actually handle the case.
Keep the facts on your pages current. In Los Angeles, Google's answer gave California's current cap correctly. The doctor-lawyer firm it named first still showed an older, lower cap in the search result for its own page. The AI can be right where a firm's own page is out of date, and it can repeat whichever it finds.
A page that explains the no. Families who've been turned down can ask Google's AI why. For the Denver family, its answer drew on firms' own pages about it.
On the Seattle map, review counts won. The three firms with the most reviews held every place. A med-mal firm's proof has to come from how it reviews a case, who does the work, verdicts it can publish, and what referring lawyers say about it.
Judge it by the signed case. The plain med-mal search looks cheap, and the cases often take years to resolve. What a channel costs per signed case is the number that tells you whether it pays.
Mid-page assessment
I ran the first family-law ad on ChatGPT, before the platform allowed legal-services ads. Before your assessment, I look at whether ChatGPT and Google's AI name your firm for the injuries you take, in your state. On the call, we map what your pages need to say first.
The questions we asked
Eleven questions a patient or family member might ask, with the city named in the question:
- New York: a mother who died after an ER sent her home with what turned out to be a stroke
- Los Angeles: permanent nerve damage after a routine surgery, and whether a case is worth it under California's cap
- Chicago: a baby born with brain damage from lack of oxygen, and how to find a birth-injury lawyer to trust
- Atlanta: breast cancer missed on two mammograms
- Phoenix: a father with stage 4 bedsores in a nursing home
- Denver: turned down by three firms, and why
- Seattle: a search for the best medical malpractice lawyer in the city
- Charlotte: a medication error at a hospital that caused kidney failure
- Kansas City: what a med-mal lawyer costs and how long a case takes
- Boise: an ER that missed sepsis, and a double amputation
- Los Angeles, in Spanish: a baby with cerebral palsy, and a lawyer who speaks Spanish
ChatGPT also got a second Denver question: a bowel injury during gallbladder surgery, after three firms said no.
Asked on October 28, 2026. We read each Google AI answer in full and asked ChatGPT one question per conversation. A firm counts as named when the answer told the person to contact it or listed it by name.
Joshua David Little is a Fractional CMO for law-firm owners and the founder of Aquila Law Firm Marketing. He has run his own consulting firm since 2004, and has done hands-on work helping attorneys prepare complex litigation for trial in family law, medical malpractice, personal injury, and unjust enrichment cases. He ran the first family-law ad on ChatGPT, before the platform allowed legal-services ads.