This AI-Fueled Startup Is Helping Attorneys Find New Class Action Lawsuits

Forbes

16 hours ago

At first glance, a 2021 lawsuit against dating app Bumble seemed like a run-of-the-mill data privacy case. The company was accused of collecting and storing Americans' face scans without their consent, which was illegal in Illinois, where the lawsuit was filed. While denying the allegations, Bumble settled the case for $40 million. Now, thousands of Bumble users are receiving about $1,900 apiece as compensation. The plaintiffs' attorneys are getting an even bigger payday, taking home $14 million.
Beyond the large payout, what makes the case unusual is the way it came about. The alleged violation at issue was discovered by five-year-old Israeli tech startup Darrow, which uses artificial intelligence to scour the internet for arguably illegal–and more importantly, class-action-lawsuit-worthy–corporate behavior. It spots the seeds of a suit in different places, such as online consumer complaints or (in the Bumble case) an app's privacy policies. Then it compares its findings to laws and regulations to gather ideas for class action lawsuits.
Darrow sells the ideas to attorneys and even helps them find people to become plaintiffs for a case through targeted digital advertising. The novel approach is a striking example of how AI is reaching remote niches of industries and transforming them (or at least speeding them up and making them more efficient). AI is already widely used by lawyers for such functions as reviewing masses of documents or emails turned over during discovery in a legal case. But the notion of using AI to discover legal causes of action is a new one. Chicago lawyer and Darrow customer Katrina Carroll, who was the lead attorney in the Bumble case, says Darrow has 'enabled me to file cases that I would never have filed were it not for their expertise.'
Cofounder and CEO Evyatar Ben Artzi, 35, sports a thick, philosopher-like beard and makes grandiose statements about Darrow's mission. He says he wants to create 'a world of frictionless justice … where you don't have to look over your shoulder to see who's screwing you … where it's easy to do business, because risk is transparent.'
It's expensive for attorneys to find and file class actions, since they require a lot of upfront research. Artzi wants Darrow to lower the cost of legal services and expand 'access to justice.' The CEO and his cofounder, Gila Hayat, 30, named the company Darrow after Clarence Darrow, the famous American lawyer who took on underdog clients such as John Scopes, a high school teacher accused of breaking a law against teaching evolution, and child murderers Leopold and Loeb.
Artzi thinks about legal risks in a very financial sense–as an asset. 'Any legal risk is a future cash flow with some kind of probability weight to it. All you need to do is surface these opportunities, give them to the experts that can handle them–whether that's a company or a law firm–and let them create the right trade, so to speak.' Beyond selling to law firms, Darrow wants large companies to buy its services so they can see their own legal vulnerabilities and fix them.
Darrow's business model is intriguing. The company makes money the same way many software firms do, by charging customers both subscription and usage-based fees. But it also has a more controversial revenue source. When lawyers win cases originated with a Darrow idea, the startup quietly takes a cut of the lucrative attorneys' fees. It does that through a little-known partnership with Arizona lawyer Don Bivens, which is enabled by a 2020 administrative order by Arizona's Supreme Court that removed an ethical rule and allowed Arizona law firms to share attorneys' fees with non-lawyers.
Two lawyers who are Darrow customers told Forbes that Darrow requires them to include Bivens as co-counsel in any case that originates from the startup's research. Bivens has full oversight over Darrow's case recommendations: He helps decide which law firms to pitch them to, reviews them before they're sent out and supervises how the advertising will work to find plaintiffs.
Darrow's chief revenue officer, Mathew Keshav Lewis, doesn't like discussing this setup. When we asked repeated questions about it, he was evasive and said Darrow doesn't disclose its 'commercial agreements.' Bivens, who passes a chunk of his attorneys' fees back to Darrow, confirmed the arrangement to Forbes . (Lewis added that Darrow's partnerships vary by jurisdiction and that some of its customers don't co-counsel with Bivens, without elaborating further.)
Lewis says his skittishness stems from a fear that competitors will copy Darrow's model, and from sensitivity around ethical rules about sharing attorneys' fees with non-lawyers. In at least 47 U.S. states, ethical guidelines prohibit non-lawyers from sharing in attorneys' fees. Some lawyers argue convincingly that the rules are outdated and protectionist. Either way, Lewis and Bivens insist that Darrow isn't doing anything improper and that the tech company has no say in lawyers' decisions on a case. They say Darrow is just providing 'legal intelligence,' or information and data to support a case, like any other consultant regularly does.
Regardless of what anti-fee-sharing sticklers might think of its business model, the 156-person startup has been growing quickly. It counts 80 law firms as customers and charges them between tens of thousands of dollars and millions of dollars a year for its services, Artzi says. The company's revenue reached $26 million in 2024, and he expects it to surpass $50 million this year. 'In 2026, we'll do $120 million of revenue,' he predicts, adding that the startup has been cash-flow positive since 2023. 'In Israel, we're the biggest legal tech company. Definitely the most profitable and revenue-generating.'
Several lawyers we spoke with are convinced that Darrow is onto something and will ultimately be one of many companies offering AI-powered lawsuit-idea services. Says Jeffrey Cunningham, a legal ethics attorney and partner at law firm Cohen Vaughan: 'I think in 10 years, this is going to be normal.' Have a story tip? Contact Jeff Kauflin at [e-mail] or on Signal at jeff.273.
A rtzi grew up in Israel, and during four years of military service, rose to become a major, leading 250 special combat operations soldiers. He went to law school afterward and became a clerk at the Israeli Supreme Court, but he was frustrated by the slow, reactive nature of the legal system and the way lawsuits originated. 'The awareness layer of the system is really limited to what people will insert in there. It's not doing anything proactive,' he says today.
In 2020, he met Gila Hayat, who had been a data scientist in Israel's elite 8200 military intelligence unit and had used AI to analyze large amounts of data. The pair cofounded Darrow that year with seed funding from Barak Rabinowitz, an Israeli venture capitalist at F2 who holds more than a 10% stake in Darrow.
Darrow cofounders Evyatar Ben Artzi and Gila Hayat Darrow AI
Today, Darrow's models uncover different flavors of possible violations, with many involving data privacy. Darrow once found out that a hospital was sharing confidential patient data, including consumers' medical conditions and illness symptoms, with online advertisers like Google and Facebook. It also regularly finds instances of employers offering poorly performing, high-fee mutual funds in their 401(k) plans, which could constitute violations of ERISA, the 1974 law that sets standards for private, employer-sponsored retirement plans.
Darrow's technology uses its own home-grown models and foundational models from the likes of OpenAI and Anthropic, tailoring them to give outsize weight to legal rules. When it recommends a class action case idea to a lawyer, Darrow estimates how many consumers were harmed, how many millions of dollars the settlement would likely fetch and its likelihood of success.
Many of Darrow's staffers are techies working in research and development who also have legal training. 'Our people are weird hybrids,' Artzi says. 'Think about the standard Israeli ex-intelligence, Unit 8200 kind-of-person who then went to law school.' The startup tries to identify 100 to 200 new cases per quarter and insists that it steers clear of suggesting frivolous cases to law firms.
It's hard to tell to what degree the revenue-sharing aspect of its business might turn customers off, hampering its growth. The ethical rules prohibiting the practice were formally adopted in the 1980s by U.S. state bar associations, whose leaders feared that non-law-firm companies cared primarily about money and weren't morally obligated to do right by the end client. Regulatory changes to these rules have been made in Puerto Rico, Washington, D.C., Washington State, Utah and Arizona, which is why Bivens has no qualms about discussing it. Fee sharing with non-lawyers is legal in the U.K. and Australia, too.
But in America, the practice is still taboo among some attorneys. One lawyer we spoke with said the revenue-sharing practice deterred him from working with Darrow.
Bivens thinks the stigma is misguided. He says revenue-sharing law firms can easily avoid ethical transgressions by making sure the business side doesn't meddle in attorneys' work. 'If I were to let a non-lawyer influence my decision about what's in the best interest of the client, I'd have to be disbarred. The lawyer has an independent duty to the client, and nobody can interfere with that.'
B oulder, Colorado-based startup Justpoint also uses large language models and AI to find lawsuit ideas, but of a different sort. Cofounded in 2018 by Victor Bornstein, who has a Ph.D. in biomedical sciences, it analyzes millions of medical records to find cases where pharmaceuticals are producing harmful, little-known side effects. Instead of class actions, Justpoint focuses on mass torts, or lawsuits where consumers are harmed to varying degrees, resulting in different payouts for different plaintiffs. Mass tort settlements are often larger than class actions. For example, decades-long litigation involving people harmed by asbestos has racked up tens of billions of dollars in settlements.
To get its piece of the juicy attorneys' fees, Justpoint takes a different approach: It has its own law firm, Justpoint Law, that is (of course) based in Arizona. So tech startup Justpoint, with its staff of 260 people, refers cases only to Justpoint Law, with its four employees, two of whom are lawyers. Bornstein says Justpoint has brought six mass tort cases involving 13,000 claims. It has only settled 'a small fraction' of those so far, with an average payout of $355,000 per claimant. Six months ago, Bornstein raised $45 million in Series A funding for Justpoint.
Artzi isn't content to stick with class actions at Darrow–he's now trying to expand into mass torts and to sell Darrow's 'legal intelligence' to large corporations. It's a hard sell, though. Companies want to know about some risks, but not all of them, because they're legally bound to fix an infraction once they're aware of it, Lewis says. A company subscribing to Darrow's service would also have to get comfortable with the awkward reality that, if Darrow finds a misdeed big enough to warrant a lawsuit, the startup might be tipping off both the offending company and a law firm that could sue it.
More from Forbes Forbes As Trump Rolls Back Federal Financial Regulation, Blue State Regulators Step Up By Jeff Kauflin Forbes How Private Equity-Owned Companies Quietly Pocket Class Action Payouts By Jeff Kauflin Forbes Everyone Hates Credit Card Disputes. This Fintech Is Using AI To Fix That. By Jeff Kauflin Forbes Forget BLS. Here's How To Take The Economy's Temperature Without Using Government Data By Brandon Kochkodin