The Robot Lawyer Settles Out of Court

A good podcast, travel troubles, and AI audio overviews

The TPA airport at night, as a reminder that if you need to actually get somewhere, fly Southwest and not Delta (or American).

A good podcast:

Cat Moon interviews the co-directors of Stanford’s Center on the Legal Profession, Nora and David Freeman Engstrom about their research on lawyer protectionism and restrictions on UPL. Did you know that AAA - not the American Arbitration Association, but the American Automobile Association - used to defend their members against car accident lawsuits and criminal cases? Me neither. From Nora:

[T]his was not about the fact that auto clubs were hurting people. In fact, in case after case where auto clubs were shut down, there was not even a shred of evidence, not even a hint that they were hurting folks, that the members weren’t happy with the services they received. Auto club membership was booming, suggesting quite the opposite, but instead, the Bar membership was worried about threats to their livelihood and the grips of the Great Depression, and they decided to just kill them. And that’s what they did, and that’s what they did. Not just to auto clubs, but to other corporate law practitioners of the day.

Give it a listen here.

About that DoNotPay Complaint…

Yesterday (as I write this) the Federal Trade Commission announced that it was taking action against DoNotPay, the “world’s first robot lawyer¹,” and that DoNotPay had agreed to a consent agreement to settle the case. This case against DNP is part of a larger crackdown by the FTC on deceptive AI claims.

Here’s each document:

Basically the FTC Complaint goes through a bunch of the DNP ad copy on their site that talks about how amazing their product is, and how it’s going to do all this amazing legal stuff for the people using it, but doesn’t. Here’s the crux of what the FTC is saying int he complaint (emphasis mine):

In connection with the advertising, marketing, promotion, offering for sale, or sale of subscriptions to the DoNotPay Service, Respondent represented, directly or indirectly, expressly or by implication, that the Service’s artificial intelligence and other technology operates like a human lawyer, including by applying the relevant laws to subscribers’ particular legal and factual situations; relying on legal expertise and knowledge to avoid potential complications, such as statutes of limitations, compensation limits, and jurisdiction, when generating legal demand letters or initiating cases in small claims court; and detecting legal violations on subscribers’ business websites and providing advice about how to fix them

To me, what DNP is really guilty of doing here is taking startup-speak and using it to mislead consumers about their product. The copy on their website is pitchdeck stuff - all-encompassing and definitive statements that tech startups use to get investors to throw piles of money at them. It’s stuff that I see all the time in legal tech, like companies saying that their legal AI “performs ‘better than human levels of accuracy’ and ‘without hallucinations.’” If you’ve worked in the legal field long enough I’m sure you’ve seen similar stuff.

It’s hard, too, because I think a lot of companies feel pressured to either make these grandiose (and false, in most cases) claims, or lose out on funding and good press. If I had a dollar for every press release I’ve seen where some legal tech company founded by biglaw associates in hoodies say they’ve solved the access-to-justice crisis, well, I wouldn’t be rich but I could probably go out to lunch for once. What’s more, no law firm wants to pony up big stacks of cash for a product that has the sales pitch of “this will mostly solve some of your problems and is accurate most of the time.” So we find ourselves in an arms race of increasingly stupid advertising claims and PR statements.

Including this image because I want to.

However I do, dear reader, want to draw your attention to the fact that two FTC Commissioners felt it necessary to issue statements about the need for legal innovation. Commissioner Holyoak’s statement, which was joined by the FTC Chair Lina Kahn, includes this gem:

For consumers to benefit from AI (as with any technology), they must be able to trust the claims that companies make about its capabilities. Importantly, this settlement does not suggest that consumers should use expensive professional services, or that companies should avoid offering innovative products that reduce the need for high-priced lawyers. The misdeeds of a few bad apples shouldn’t dampen pro-consumer innovation. Indeed, we are hopeful that AI will give consumers access to many types of services at lower cost and with greater convenience than has previously been available.

And Commissioner Ferguson said this:

I write also to clarify that my vote should not be taken as support for the State Bar of California’s claim that DoNotPay was engaged in the unauthorized practice of law. The Commission does not enforce state occupational-licensing laws like California’s unauthorized practice-of-law prohibition. And if a company were to create a computer system capable of giving accurate legal advice and drafting effective legal documents, or honestly advertise a system that provides something less, I doubt that the aggressive enforcement of lawyers’ monopoly on legal services would serve the public interest.

Couldn’t agree more.

AI Audio Overviews in Google’s NotebookLM

If you’re not familiar with Google’s NotebookLM product, and you have a google account, you should probably go check it out. It’s honestly one of the best RAG-esque applications I’ve seen out in the real world. It’s still ‘experimental’ which means it’s free but the G Team could shut it down at some point in the future.

Basically you create “notebooks” where you can upload sources, and then ask questions or tell it to create outlines of those sources. If I was in law school now, I’d be uploading cases from Google Scholar to it and having it create outlines and summaries for damn sure.

They recently added a feature that lets you create an “audio overview” of the materials in a notebook, which is basically a two-person podcast style discussion. It’s frankly astonishingly good. Here’s one it made from the Florida Bar v. TIKD case:

You don’t really have any control over what the podcasters talk about, other than the materials you provide. But you can imagine this type of knowledge delivery being really interesting for explaining different legal topics. And maybe we can expect a flood of ai-generated podcasts, and then tiresome AI critics on Bluesky saying how this is the death-knell of the podcast industry as a treat!

Travel travails:

Last week I was supposed to present on AI and regulation with Natalie Knowlton at the SRLN conference in Salt Lake City, and then go to ATL to present on AI and - you guessed it - regulation at the Federation of Associations of Regulatory Boards conference. But only one of those two things happened, thanks to the magic of Delta Airlines, so I completely missed the SRLN conference, which I’m still bummed about. Next time I’ll fly Southwest.

1. Frankly this is stolen valor from “Robot Lawyer LISA” which billed itself as “the world’s first impartial robot lawyer” back as far as 2016. Sadly the link to try LISA no longer works.