ContractsProf Blog

New York’s Attorney General Forces Doordash to Pay Workers for Stolen Tips

NY Attorney General Letitia James

According to Andy Newman, writing last February in The New York Times, in the earlier years of Doordash, the company encouraged customers to tip and assured them, “Dashers will always receive 100 percent of the tip.” For two years, Doordash guaranteed its delivery people (“Dashers”) $7/delivery. But if a customer tipped $3, the company would still pay the Dasher $7 and keep the tip for itself. It’s so petty and cruel, it’s hard to imagine the mindset that hatched this plot. 

Mr. Newman first reported on the practice back in 2019 when Doordash discontinued it in the face of wholly understandable customer outrage. At the time, Doordash had a market valuation of $7.1 billion. Stealing $3 tips from the workers at the core of its business model is just disgusting. Instacart, Mr. Newman reported, had a similar policy.

In February, Doordash settled with New York’s Attorney General’s office and agreed to distribute $16.8 million to 63,000 workers, some of whom could received as much as $14,000. This follows similar settlements in Illinois and Washington D.C.  Doordash was non-committal when asked if it would enter into voluntary settlements in other states.

I wish I used Doordash so that I could stop using Doordash. Punitive damages seem appropriate here. I’m not outraged only on behalf of the employees from whom a multi-billion dollar business stole tips. I would be outraged as a customer to know that tip money I wanted to give to a delivery person when to a multi-billion dollar business instead. In addition to thinking that the money should be transferred from the business to the workers, I also want restitution. Doordash has had that money to play with for six or eight years. New York’s Attorney General’s office estimates that New Yorkers placed over 11 million orders while the policy was in effect. Unless New Yorkers are very skimpy tippers, and I don’t believe that they are, $16.8 million is a very low estimate of the amount of money the company misappropriated. 

And it’s not as if there is a simple solution, like just use UberEats. As discussed here, one order from UberEats may subject you to their terms of service with respect to all related companies indefinitely.

Belatedly Introducing the Money Stuff Podcast, with Matt Levine and Katie Greifeld

A few years back, I started following Matt Levine (left) and his Money Stuff column on Bloomberg. His columns inspired posts such as this one about a costly but understandable error at Citibank and this one about a preposterous money laundering sting.

I recently discovered that Mr. Levine has taken to podcasting about his Money Stuff column, along with his co-host Katie Greifeld, (below right), who provides an intelligent conversation partner for Mr. Levine. There is a bit of a yin/yang thing between them, in that Matt likes markets to work efficiently and Katie likes chaos. But both like humor, and that is what makes Money Stuff not just educational but fun. I have gone back to the first episode and am catching up. I expect that I will have a lot of blog fodder from the podcast, as I learn so much form it and it gives me so many fresh ideas to write about. 

I should caution that, unlike Matt Levine and Katie Greifeld, I am no expert on money stuff. They do an amazing job making complex financial transactions understandable. But almost every segment inspires a short sidebar from the perspective of a ContractsProf who taught Business Associations for ten years. Moreover, being a year behind them, I can provide updates to stories that have now receded from public attention.

The first episode has four segments, each of which provides blogworthy content. The first segment is about the Destiny Tech 100 Fund (DXYZ). Destiny gives investors the ability to purchase an interest in shares of privately-held tech companies. As of their most recent disclosure, about half of the portfolio is SpaceX. Money Stuff focuses on the odd fact that when Destiny went public, the valuation of the stock mushroomed to $500 million when the valuation of the portfolio was only $50 million. When the Money Stuff folks did their reporting one year ago, the stock had just peaked at $60/share. It then fell back to earth, mostly hovering between $10-15/share, but then it exploded again after the elections to $70/share. It’s now settled in at around $30/share.

One recurring theme on the podcast that I want to highlight is rational irrationality. Given the subject matter, it should not surprise that my observations are derivative [pause while you appreciate that joke]. I make no claims to original insights for the ranks of the professoriate, but some readers of this Blog may not think about such matters very often, so my rudimentary takes may be new to them or they may be a helpful reminder.

I thought that last week was a big week for rational irrationality. There was no real good financial news in the media, but the markets exploded on rumors of good news on the horizon. A rational person would not put much faith in this administration's promises of new trade deals in the works or relaxations of tariffs rates. But the markets are not rational; they believe what they want to believe, so the rational money buys on the dip and will no doubt dump when the markets whipsaw with the next battle in the trade war.

My two cents to add on to this segment is that there are two possibilities. One is that these privately held tech start-ups are radically undervalued, which I doubt. The other is that a lot of people want in on these companies, and they are willing to pay far in excess of the companies' valuation in order to do so. Economic theory tells us that stock prices should not respond to laws of supply and demand. If the value of a stock is $10/share, it should sell at that price, no matter how many people want to own it. But investors can be irrational. Rational investors know of that irrationality and are happy to exploit it, boosting the stock price well above the value of the underlying asset and then selling to the irrational marks once the stock has peaked.

I did a quick search to see if more has been written about Destiny, but as is typical with such things, there was a flurry of reporting when the stock price popped about a year ago, and since then, mostly silence. Finance sites follow the stock price, of course, but I couldn’t find much analysis of the fund as a phenomenon. 

The second segment is about Avi Eisenberg, who devised an incredibly simple trick for making $117 million in 2022 by manipulating positions (longs and shorts) in Mango Markets (MNGO), a decentralized finance platform. Matt Levine’s money quote [I pause again while you appreciate the joke] describes both this caper and the attitude of crypto capitalists generally. Matt and Katie focus on the tension between the regulators’ perspective that  “law is law” and the crypto-investors perspective that “code is law.” Matt says, “As long as you’re the smart one, you’re like ‘Ahhhh, anything goes, you don’t need regulators,’ but then you get blown up and you’re like, ‘Ahhhhh, where were those regulators?” Worth a listen just to hear Matt’s “Ahhhh, which I can’t really translate into print.

Mr. Eisenberg was so confident that code is law that he took to Discord to brag about his “hack” and to defend it as completely consistent with the protocols that Mango had set up. Those protocols had a vulnerability; he exploited it. Things went a bit too well. Mango’s losses exceeded its insurance coverage, and the platform was in danger of collapse. Mr. Eisenberg agreed to return $67 million of his winnings in exchange for Mango’s promise not to press charges.

Well, federal prosecutors don’t think that code is law. They think law is law, and as Amin Ayan reports here on Cryptonews, prosecutors are now seeking a 6.5 year sentence for Mr. Eisenberg, who was convicted in April 2024 for wire fraud, commodities fraud, and market manipulation. I find the wire fraud and commodities fraud charges a bit puzzling, as one of the interesting features of this story is how open Mr. Eisenberg was about what he was doing. Market manipulation may be another matter. Having looked at the indictment, I see that he was charged under 7 USC § 13(a)(2), which makes it a felony for:

Any person to manipulate or attempt to manipulate the price of any commodity in interstate commerce, or for future delivery on or subject to the rules of any registered entity, or of any swap, or to corner or attempt to corner any such commodity or knowingly to deliver or cause to be delivered for transmission through the mails or interstate commerce by telegraph, telephone, wireless, or other means of communication false or misleading or knowingly inaccurate reports concerning crop or market information or conditions that affect or tend to affect the price of any commodity in interstate commerce, or knowingly [to violate various other provisions of the Code.]

The unusual thing about market manipulation is that it doesn’t require any deception, and it seems to apply on the facts here.

According to Alex Costa, reporting on DailyCoin, Mango Markets shut down in January, never having recovered from having its vulnerabilities exposed. Mango is seeking restitution of the remaining $47 million that Mr. Eisenberg kept from his scheme. He’s got other problems, as prosecutors found child pornography in his possession while investigating his other crimes. 

Very few sectors of the economy are as risk-averse as academics. Law professors and perhaps business professors like to tell ourselves that we could have made a lot more money if we had stayed in private practice, and that is certainly true for some of us. But most of us have happily traded job security, status, and the ability to control our own lives for the pursuit of wealth. I admit, I don’t think I could have made huge sums as an attorney, because I could never work whole-heartedly on behalf of clients to whose fate I was largely indifferent. I lasted as long as I did because I always, with one exception, esteemed my client’s adversary somewhere below indifference. Being averse to risk and not really interested in vast wealth, I have a hard time understanding people who devote their lives to making something from nothing.

At my former law school, I had colleagues who viewed life from the perspective of a tax attorney looking for the lucrative loophole. They had something like mild contempt for people who were governed by norms rather than by law. And now an infinitely more extreme version of such people runs the country. We need to appreciate thee psychology and the worldviews of these people and do so fast.

Reefer Brief: Fallout from Alternative Payment Partnership Collapse

It is an exceedingly rare thing that I read a case for our Reefer Brief feature, and I think, “That’s a really good idea!” But that is the case here. As regular Reefer Brief readers (if there are any) already know, even though states have relaxed regulation of marijuana, the federal government has not done so. As the American Bankers Association puts it:

Currently, thirty-seven states, the District of Columbia, Guam and Puerto Rico have all legalized the use of marijuana to some degree. Yet the possession, distribution or sale of marijuana remains illegal under federal law, which means any contact with money that can be traced back to state marijuana operations could be considered money laundering and expose a bank to significant legal, operational and regulatory risk.

As a result marijuana businesses face challenges when it comes to financial transactions. It is hard for them to maintain bank accounts, and so a lot of the business operates on a cash basis. 

Christopher Rentner saw an opportunity. He founded a cannabis financial technology startup, Plaintiff Emerging Industry Technologies, Inc., called “Spence” in the opinion, which aimed to provide “compliant [cash] payment alternatives to legal dispensaries and merchants through secure and transparent electronic payment platforms.” The idea was to work with banks to allow for electronic transfers and thus to enable marijuana businesses to keep less cash on hand.  Spence also introduced a Buy Now Pay Later feature unique in the marijuana industry.

Enter defendant Fidelity National Information Services (FNIS). Just as Spence was getting off the ground, FNIS, through its $43 billion acquisition of Worldpay, became the world’s largest processing and payment company. FNIS at first wanted to acquire Spence, but instead the parties negotiated a collaboration. FNIS drafted a purported Partnership Agreement. FNIS was to invest $4.5 million in Spence, and Spence was to facilitate the use of Worldpay in the marijuana industry, working exclusively with FNIS and setting up its systems to coordinate with the Worldpay platform. The partnership was to have an initial five-year duration after which it was renewable in two-year increments.  FNIS promised to “take care” of Spence, but it also demanded a great deal from Spence, including a large equity stake. Spence spent millions of dollars on recruiting fees, legal fees, development fees, salaries, and hardware purchases to comply with FIS’s requirements.

“We’ll take care of you” seems have become a magical phrase that well-resourced parties know will induce weaker parties into cooperation but courts find too vague to be enforceable.  It worked for David Chase; it worked for the L.A. Clippers. Would it work here as well?  FNIS kept its investment in Spence on the down-low, preferring to characterize the parties’ relationship as a partnership, but it also gave Spence private assurances, leading it down the old primrose path.

Meanwhile, FNIS was having difficulties. It sold a majority stake in Worldpay to a private equity firm. The story is complicated, but cutting to the chase, FNIS backed out of its partnership with Spence and then proceeded with its own cannabis financing scheme. Having lost its partner, Spence also lost all that it had invested, in terms of personnel and resources, in the partnership. It soon shuttered its operations. It looked like an FNIS subsidiary was going to purchase Spence’s assets, but FNIS then told Spence that it would vote against the deal unless Spence agreed to waive all its legal claims against FNIS. When Spence refused to do so, FNIS made sure that nobody could purchase Spence’s assets and compete with FNIS in the marijuana financing space. It’s diabolical!

Spence sued for breach of contract, fraud, unjust enrichment, and promissory estoppel. FNIS moved to dismiss. In August, the U.S. District Court for the Northern District of Illinois issued its ruling on that motion in Emerging Industries Technology, Inc. v. Fidelity National Information Services, Inc. The court begins by assessing whether the purported Partnership Agreement mentioned above is in fact an enforceable agreement. The court spends some time highlighting the gaps in the document and concludes, "The so-called partnership agreement created an outline for a potential business deal. But that’s about it.” Too many material terms are left to be determined. FNIS’s public announcements of a partnership are also insufficient. There must actually be a partnership, and here there is no evidence that one was formed. The court dismissed Spence’s claim for breach of contract.

Spence’s fraudulent concealment claim at first seemed to fair no better. On one theory, the claim required  a special relationship giving rise to a duty disclose, and here Spence’s claim of a special relationship seemed to turn on the parties’ disparities in sophistication and bargaining power. The court, which is given in this opinion to drifting into casual formulations sums up Illinois law on the subject as follows: “A special relationship does not exist merely because Joe Schmo enters into an agreement with Big Company.” Moreover, Mr. Rentner, Spence’s principal was no inexperienced rube. However, Spence had made sufficient allegations, just barely, to support its claim of fraudulent concealment. It had alleged that FNIS had failed to correct some of its misleading statements.

Spence’s promissory estoppel claim also barely remained alive. Spence had alleged just enough so that the court was unable to rule out the possibility that it had reasonably relied on FNIS’s statements, including representations about sharing future profits and FNIS’s public announcement of a partnership. For the same reasons, Spence's claim for fraudulent inducement survives. The unjust enrichment claim also survives, pending Spence’s ability to make out a claim for either promissory estoppel or fraudulent inducement.

Court Finds Twitter’s Terms of Service Unconscionable in Part

Plaintiffs allege that Twitter was a dumpster fire. Twitter’s Head of Security from 2020 until 2022 turned whistleblower and testified to Congress about pervasive problems with Twitter’s data security. Plaintiffs allege that, due to Twitter’s negligence on that front, Twitter experienced a massive data breach, and Twitter users’ personal information was harvested and then sold on the dark web. Plaintiffs allege special harm because they took advantage of Twitter’s invitation to users to post under pseudonyms. The data breach made it possible for people to establish the identities behind their posts. They filed a class action complaint, alleging seven causes of action, including breach of express and implied contracts. The others are of less interest to us beyond the fun bit where Twitter moves to dismiss the claim for gross negligence despite having already conceded that California law forbids limitations on liability for gross negligence. The court dryly notes that "it is unclear why Defendant would raise this argument here."

In Gerber v. Twitter, Inc., the issue was the enforceability of Twitter’s Terms of Service (ToS), which put Twitter’s users on notice that its services were provided AS IS and without warranties and that Twitter’s liability was limited to the maximum extent provided by law. Kandis Westmore, Magistrate in the U.S. District Court for the Northern District of California got right to the heart of the matter, noting that ToS are enforceable unless unconscionable.

Data Breach, Image by Microsoft Copilot

California tests unconscionability on a sliding scale, requiring some combination of procedural and substantive unconscionability. The court found that the ToS were “at least somewhat” procedurally unconscionable, in that they are a form contract of some length, and the objectionable terms appear on pages eight and nine of a twelve-page document. Twitter objected that the language was conspicuous, in large font and ALLCAPS and that it provided notice to users each time it updated its terms. Still, the court noted plaintiffs’ objections that "these terms were buried in lengthy forms drafted by the party who wished to enforce them.” In California, that is enough to establish at least some procedural unconscionability.

As to substantive unconscionability, while parties can disclaim liability, the problem here is that, taking the allegations of the complaint as true, Twitter had statutory duties to protect its users against data breaches and knowingly failed to do so. The court rejected Twitter’s claim that the statute in question cannot be relied on in support of common law claims for breach of contract or negligence. California courts have found otherwise.

The court granted Twitter’s motion to dismiss Plaintiffs’ allegations of a breach of an express contract. Plaintiffs relied on blog posts and website statements that they did not adequately link to the User Agreement. Moreover, Plaintiffs conflate Twitter’s promise not to disclose users’ personal information without their consent with a failure to maintain adequate data security measures. As a result, Plaintiffs could identify no express promise that Twitter breached. However, Plaintiffs did successfully allege breach of an implied contract based on representations on Twitter’s website about its commitment to data security. Those promises were then breached if, as alleged, Twitter failed to take steps to safeguard users’ information. I am not sure why the court dismissed Plaintiffs’ express contract claims but then upheld “implied” claims that are based on written promises. A written promise seems like an express promise to me.

In the end, the court dismissed plaintiffs claims for breach of an express contract and denied Twitter’s motion with respect to all other claims.

Tip of the hat to my former student Don Dechert, who alerted  me to a post about the case on Eric Goldman’s Technology and Marketing Law Blog. Professor Goldman notes that the ruling is quite broad, rendering ToS ineffective to shield companies for liability for intentional conduct. There is no clear way to fix that infirmity. One might suggest that sophisticated technology companies not knowingly fail to protect their users from data breaches, but of course all plaintiffs have to do is plausibly allege knowing misconduct to create a litigation headache for the defendants.

Twitter's New Liquidated Damages Clause

Image by DALL-E

We recently posted about Twitter's venue clause in its latest Terms of Service (ToS), which went live on November 15th and can be found here. There is another aspect to the new ToS that has gained some notoriety. It is Twitter's new liquidated damages provision, which provides as follows:

Liquidated Damages

Protecting our users’ data and our system resources is important to us. You further agree that, to the extent permitted by applicable law, if you violate the Terms, or you induce or facilitate others to do so, in addition to all other legal remedies available to us, you will be jointly and severally liable to us for liquidated damages as follows for requesting, viewing, or accessing more than 1,000,000 posts (including reply posts, video posts, image posts, and any other posts) in any 24-hour period - $15,000 USD per 1,000,000 posts. You agree that these amounts are (i) a reasonable estimate of our damages; (ii) not a penalty; and (iii) not otherwise limiting of our ability to recover from you or others under any legal or equitable theory or claim, including but not limited to statutory damages and/or equitable relief. You further agree that repeated violations of these Terms will irreparably harm and entitle us to injunctive and/or other equitable relief, in addition to monetary damages.

There is a lot going on here, both legally and in terms of the back-story behind this provision. 

Calling something a liquidated damages provision does not mean that it is enforceable. Courts will look at the provision and determine on their own whether it is an unenforceable penalty clause. Stating in more detail that the counterparty agrees that the clause is not a penalty should not really change anything. Even in negotiated agreements, courts undertake their own assessment of whether a provision is a penalty. One thing the court might consider is sub-point iii, which provides that Twitter can recover damages on top of liquidated damages. That undercuts the advantage of a liquidated damages provision -- saving litigation costs by stipulating to damages in advance -- and thus suggests that this clause, notwithstanding its insistence to the contrary, is in fact a penalty.

Mike Masnick, writing on Techdirt, provides the best explanation I have seen of what motivated the new liquidated damages provision. Elon Musk may be trying to set up a suit against AI companies, including his recent nemesis OpenAI, that scrape massive amounts of information from websites like Twitter. Mr. Masnick provides an additional reason why the provision might be unenforceable, at least in part.  Section 40.12 of the European Union's Data Security Act requires platforms like Twitter to provide access to their data:

Providers of very large online platforms or of very large online search engines shall give access without undue delay to data, including, where technically possible, to real-time data, provided that the data is publicly accessible in their online interface by researchers, including those affiliated to not for profit bodies, organisations and associations, who comply with the conditions set out in paragraph 8, points (b), (c), (d) and (e), and who use the data solely for performing research that contributes to the detection, identification and understanding of systemic risks in the Union pursuant to Article 34(1).

Mr. Masnick thinks the liquidated damages provision might be an attempt by Musk to poke the EU regulatory bear, but it is also possible that Musk (or his attorneys) understand that there has to be a regulatory carve-out from the rule. In that case, the real target of the provision would be private scrapers.

Thanks to Mitu Gulati for sharing news of this controversy with me and for sharing with me this interesting discussion of developments in liquidated damages law by Glenn West.

Massachusetts Supreme Judicial Court Upholds Uber's Terms of Service & Compels Arbitration

William Good, a Boston chef, took an Uber home from work in April 2021. During that ride, he was involved in an accident and sustained a neck injury that left him a quadriplegic. Five days before that catastrophic Uber ride, Mr. Good had use the Uber app, which he had been doing since 2013. This time, before proceeding to the app, Mr. Good was confronted with a blocking screen (left) that notified him that Uber had updated its terms of service and required his electronic agreement to the new terms, which he was invited to view via a hyperlink. Mr. Good, because such is the way of all flesh, checked the box at the bottom. The terms of service included an arbitration clause and insulated the company in various ways from liability for the actions of its drivers. 

After his injury, Mr. Good sued the driver and Uber. Uber duly moved to compel arbitration, but the trial court denied that motion on the ground that Mr. Good had no reasonable notice of its terms and had not manifested assent to them. 

In June, in Good v. Uber Technologies, Inc., the Supreme Judicial Court of Massachusetts reversed, and ordered the trial court to grant Uber's motion to compel arbitration. The opinion is extraordinarily lengthy, and there was also a lengthy dissent.

The Court begins by noting that Mr. Good did not have actual notice of the terms to which he agreed. The Court cites scholarship establishing that consumers do not read terms of service. The Court then stressed the importance of reasonable notice of terms in situations such as this one, where the service being contracted for is not particularly costly and one would not expect complex terms. Notwithstanding the exacting standard for notice, the Court found that Uber had provided sufficient warning to put Mr. Good on reasonable notice of its terms. The hyperlink to Uber's terms of service was conspicuous. Just one click would have taken him there, and then he would only have had to read five paragraphs into the document to learn of the arbitration clause. Mr. Good had to consent to the terms in order to proceed to order a ride.

Mr. Good argued that the Court ought to require more. The Court could require scrollwrap, a version of wrap contracting in which the user has to scroll through the terms of service in order to get to the box indicating assent to terms at the end. In the alternative, it could require that users be forced to actually click on the hyperlink before they can be said to have assented to terms. According to the Court, to do so would be to require actual assent. The law requires only reasonable notice of terms. Nor would the Court require that vendors highlight key terms to enhance the users' notice of terms that they would not read in full. The duty to read governs even in situations where we know that very few people read.

The dissent maintains that, "This case is not in any way ordinary, as the court contends." Mr. Good thought he was just signing up for a ride. He had no way of knowing that Uber, a sophisticated technology company, was disclaiming all liability for any injuries he might sustain during that ride. I think that is absolutely true, but I also think that it is perfectly ordinary, at least in the United States, for powerful business entities to use form contracts to escape liability for tortious conduct by the entities or their agents/employees/independent contractors. The dissent thinks that the law here is unsettled. I wish I could say that I agree, but we have reviewed multiple cases in this space that look an awful lot like this one.  There is an outlier case from Pennsylvania, Chilutti, but the Pennsylvania Supreme Court is reviewing that case, and I suspect that it will reverse. Stay tuned. 

This Bird Has Flown: Leaving Twitter

This move is long overdue. I will permanently close the Blog's Twitter account at the end of the week.  I really enjoyed my time curating the Blog's profile on Twitter. I learned a ton from that site and made/solidified some relationships. However, these days, the Blog gets very little engagement through Twitter. It's a crowded marketplace, and the tone of the Blog is not ideally suited for that space.

If you would like to continue following the Blog on social media, we will maintain our Mastodon and Bluesky accounts. I recommend both sites. They operate a lot like Twitter, but somehow the vibe is very different. 

Bluesky is more like Twitter. It was founded by two Twitter executives. I don't really know or care what caused two of the people behind Twitter to create a rival site that is a lot like Twitter. All I an say is that Bluesky is growing very rapidly now, as there has been a recent exodus from Twitter. The number of users was 200,000 in July 2023, 5.9 million in July,2024, and I recently saw that it surpassed 14 million this month.  You used to need an invitation to join Bluesky. That is no longer true. One of the problems I experienced with Twitter , especially this year,was that most of the Blog's followers seemed to be bot accounts. That seems to be very rare on Bluesky. Lately, Bluesky has been featuring a lot of "starter packs" which helps new users find people with common interests and also helps existing accounts get a lot of new followers who share common interests. The Blog's followers have nearly doubled just in the past week, and engagement is way up (although the numbers are still very small).

Mastodon seems to attract more tech people and visual artists. Photos are much clearer and crisper on Mastodon. It's a very pretty site. There's a lot of happiness and support on the site -- at least on the part of it that I inhabit. The downside (for me) is that it attracts a lot of techies who, for reasons that escape me, have no interests in contracts law. Also, their posts are completely incomprehensible to me. 

I recommend both of these sites, and I hope that those of you who follow the Blog on Twitter will continue to do so on one or both of these sites.

California District Court Grants Preliminary Injunction to Pepperidge Farm Distributor

This case illustrates what happens when COVID-19 and the need for ready access to Pepperidge Farm Goldfish crackers collide. Train wreck. You can't look away.  Pepperidge Farm's behavior in this litigation certainly is fishy, and the District Court doesn't seem inclined to catch and release.

In August 2017, Vital Distributions, LLC (Vital) entered into an agreement with Pepperidge Farm, Incorporated (Pepperidge) to serve as its exclusive distribution agent for two California counties.  At the time the parties entered into their agreement, Pepperidge also proffered an E-Commerce Acknowledgment (the Acknowledgment), which purported to put Vital on notice that the parties' agreement did not prevent Pepperidge from entering into separate distribution agreements through electronic means.  At the time that the parties executed their agreement, Vital rejected the Acknowledgment and refused to sign it.  Vital made clear its understanding that it was to serve as Pepperidge's exclusive distributor for all purposes in the two California counties.

Vital was aware of an Amazon distribution center in the region.  It knew that its agreement with Pepperidge would be nearly worthless if it had to sign the Acknowledgment.  Pepperidge's representative warned that Pepperidge might not agree to sign without the Acknowledgment, but in the end Pepperidge did sign.  Under the agreement, Vital got paid for storage and delivery services, and it also received commissions on all distributions, whether through ordinary or e-commerce, within its territory.

The agreement included a carve-out for deliveries to chain stores that insisted on direct deliveries from Pepperidge to their warehouses, but only after Pepperidge attempted in good faith to persuade the chains to use Vital's services and those chains insisted on bypassing the agreement.  Finally, Pepperidge was contractually obligated to provide amounts of its products to Vital sufficient to guarantee an adequate and fresh supply.

The parties worked together well until COVID hit. Then, product panics hit. Demand for Pepperidge products soared, especially Goldfish crackers, and Pepperidge started meeting that demand through Amazon. Even after shoppers stopped braining each other in pursuit of toilet paper, Pepperidge continue to meet demand through Amazon while providing Vital with half the product it had previously supplied, harming Vital's business relations with retail stores. Vital's customers took to ordering the product they so desperately needed through the Internet.

Vital sued alleging breach of contract and breach of the implied covenant of good faith and fair dealing.  It sought an accounting and declaratory relief.   Pepperidge's motion to dismiss was denied; Vital's motion for discovery was granted, as was its motion for a temporary restraining order.  Back in April, in Vital Distribution, LLC v. Pepperidge Farm, Inc., the District Court granted Vital's motion for a preliminary injunction.

Supervillain Snidely Whiplash, auditioning for the role of Pepperidge Farm in the film version of this case

Soon after an unfavorable discovery ruling, Pepperidge Farm sought to exercise a buyback option provided for in Article 20 of the original agreement.  Article 20 permitted Pepperidge Farm to buy back the distributorship for its market value plus 25%.  Charmingly, it exercised this purported option through an e-mail sent by a paralegal providing Vital with thirty minutes notice. It then disabled technology essential to Vital's distribution services. These actions were then subject to a temporary restraining order.  Pepperidge represented that the parties were working out a settlement that would effectuate the transfer of the distributorship back to Vital. In fact, Pepperidge continued to engage in shenanigans undermining Vital's relationships with distributors. Pepperidge attempted to characterize its conduct as business as usual.  The parties' relationship had soured and so it was exercising its buyback option.

The Court had little difficulty concluding that Vital would likely succeed on the merits of its breach of contract claim. Pepperidge did itself no favors by refusing to respond to recovery requests, making it easy for the Court to side with Vital on all material facts that might otherwise be disputed.  Similarly, the Court found that Vital would likely succeed on its claim that Pepperidge invoked the buyback provision in violation of the duty of good faith and fair dealing. Pepperidge presented no evidence of inadequate performance by Vital and thus had no justification for its invocation of the buyback provision or for its curious timing.  Bad faith is an amorphous concept, the Court noted, but we know it when we see it. If something looks like a fish, crunches like a fish, and tastes of cheddar, well, it's probably a Pepperidge Farm Goldfish being sold through e-commerce distributors in violation of an exclusive distribution agreement.

As you can imagine, the irreparable harm analysis, balance of the equities, and public policy considerations all went Vital's way.  The Court granted Vital's motion preliminarily enjoining Pepperidge from terminating or buying back Vital's distribution rights.

Ninth Circuit Reverses Summary Judgment and Remands Case Alleging Privacy Violations by Google

Speaking of scraping, well this isn't exactly a scraping case, but it's definitely scraping-adjacent.  Plaintiffs sought to certify a class consisting of Google Chrome users who sought to avoid data collection by choosing not to sync their browsers to the Google accounts.  Their belief that Google would not collect their personal data was based on Google's "Chrome Privacy Notice."

The District Court denied Google's consent-based motion to dismiss, but after an evidentiary hearing, it granted summary judgment to Google based on its finding that data was shared with Google on a "browser-agnostic" basis.  That is, data flowed to Google even if web browsers were not using Chrome. Plaintiffs, on notice of Google's general data-collection practices, could not complain that the same data collection occurred through their non-synced Chrome accounts.

In August, in Calhoun v. Google, LLC, the Ninth Circuit reversed.  Google's general privacy policy pertains to IP addresses and cookies. The Chrome-specific privacy policy related to broader categories of data, including "browsing history, bookmarks, tabs, passwords and autofill information, and other browser settings." The Ninth Circuit chided the District Court for holding a 7.5 hour evidentiary hearing that got sidetracked on browser agnosticism and failed to ask the legally pertinent question: was a reasonably prudent web browser on notice that by using Chrome in the non-synced setting, they were consenting to the collect of their personal data.  At this point in the proceedings, plaintiffs had sufficient alleged that the notices Google provided were insufficient.

The case was remanded to the District Court with instructions to apply the appropriate standard.  Only then can the District Court make a determination of whether a class should be certified.

Even Judge Easterbrook Won't Enforce Ancestry.com's Arbitration Agreement Against Minors

Plaintiffs in this case are children. Their guardians registered on Ancenstry.com and in so doing agreed to arbitration.  While the plaintiffs' guardians sent in the plaintiffs' saliva samples to Ancestry.com, plaintiffs did not read Ancestry.com's terms, nor were they required to do so.  Plaintiffs allege that they never created their own Ancestry.com accounts, did not access their guardians accounts, did not receive their DNA test results, or interact with Ancestry’s website in any way before filing suit. 

Ancestry.com was then acquired by Blackstone, Inc. (Blackstone), and plaintiffs allege that Ancsetry.com violated their privacy rights by disclosing private genetic information to Blackstone. Ancestry.com moved to compel arbitration. The district court denied the motion, and in Coatney v. Ancestry.com DNA, LLC, the Seventh Circuit affirmed. The opinion is by Judge Brennan, but in fairness to Judge Easterbrook, I note that he joined the opinion, causing me to further revise my view that his position is "arbitration for all."

Image by DALL-E

On appeal, Ancestry.com first argued that plaintiffs were bound by its terms through the conduct of sharing their DNA. The Seventh Circuit rejected that argument because the terms of Ancestry.com's agreement with plaintiffs' guardians unambiguously binds only the guardians. Ancestry.com cited only one unpublished district court opinion in support of its consent-through-conduct argument, but the Seventh Circuit found the case distinguishable, based on the very different language in the relevant agreements.

Ancestry.com next tried to argue that plaintiffs were bound as closely-related or third-party beneficiaries. This section of the opinion is very lengthy, but the Seventh Circuit states right from the start that Illinois law has a strong presumption against biding third-parties to contractual obligations. To make matters worse, Ancestry.com's terms exclude third parties.

Ancestry.com's final argument is that plaintiff's directly benefitted form the guardians' accounts and thus are estopped from challenging the motion to compel arbitration.  This section of the opinion is also very long, in part because there is scant authority for the doctrine of direct-benefits estoppel under Illinois law.  I wonder why courts don't more often certify questions to the state supreme court in such instances.  Instead, the Seventh Circuit plows ahead.

The fact that the plaintiffs in this case never accessed Ancestry.com themselves or saw any of the genetic information that their guardians gathered makes this case much easier.  It also likely makes the ruling very narrow.  My instinct, based on zero experience with Ancestry.com, is that it would be rare that parents would gather their children's genetic material and send it in to Ancestry.com for analysis and then that the family would never discuss the results.  If my instincts are correct, this rare victory for the foes of mandatory arbitration may be limited to its unique facts.

What's All the Fuss About? The Great Scrape

Occasionally, new private law scholarship posted on SSRN gets downloaded by thousands of people.  When it does, inquiring minds want to know what all the fuss is about.  This feature of the blog gives you the tl;dr on what you really ought to be reading for yourself.  Today's subject is the most recent paper by Daniel Solove (below left) and Woodrow Hartzog (below right), The Great Scrape: The Clash Between Scraping and Privacy, which is pushing 2000 downloads on SSRN.

Scraping, the Authors tell us is. the automated extraction of large amounts of data from the internet.  Through scraping, actors gather enormous amounts of data and personal information (worrisome) without notice or consent (troubling), and then this information provides fodder for AI tools such as facial recognition, deep fakes, and large language models (panic-inducing). (4 - parentheticals added).  The Authors concede that scraping has its socially beneficial uses, but scraping of personal data "violates nearly every key privacy principle embodied in privacy laws, frameworks, and codes" and is, in short, "antithetical to privacy." (4) While scrapers contend that they make use of publicly available data, courts have recognized a privacy interest in publicly-available but practically obscure personal information. (4)

We need scraping to have a useable Internet, but scraping is in fundamental tension with basic privacy law.  The Authors call for responding to the Great Scrape with the Great Reconciliation of scraping and privacy norms. (5)

Part I of the Article provides a history and explanation of scraping.  We first learn that scraping, that is, online data harvesting, has been around as long as the Internet (7-9), but the power of scraping tools has grown vastly in the age of AI. (9-10) If you are on this site, statistically, it's more likely that you are a bot scraping the blog than a human reading the blog.  Now, if you happen to be a bot, I'm not judging you.  The Authors say I can't because the scraping of personal data occurs in the murk of an ethical twilight zone. (11-13) Which brings us to the current conundrum of "scraping wars."  Some of the very websites that hire scrapers to enhance their functionality now object to being scraped for other purposes. (13-14) They are fighting back against the scraper through legal challenges with theories ranging from trespass and fraud to business torts and violations of privacy protections, (14-20) and by trying to use technology so that they can fight fire with firewall. (20-21)  While scrapers are trying to buy out the resistance (23), regulatory intervention might change the market conditions for doing so. (23-27) The Authors highlight EU regulatory actions against Clearview AI. (25-26) While the FTC may have the legal means to regulate scrapers, it is not clear that it has the political clout to do so. (26-27)

In Part II, the Authors detail the fundamental tension between scraping and privacy.  Privacy law is governed by bedrock principles known as the Fair Information Practice Principles (FIPP).  FIPP comes down to  three rules: only collect data when necessary, keep the data safe, and be transparent.  According to the Authors, scraping violates all of these principles. (29). The overarching goad of FIPP is fairness, but the Authors also list seven other fundamental principles. (30-38). Their conclusion is not optimistic: "It is not clear that scraping can be performed in a privacy-friendly way." This is so because both the fundamental principles of privacy and the building blocks of privacy laws are "in dramatic conflict with scraping." (38)

Scrapers defend themselves by claiming that they only access publicly available information.  In the next section of their paper, the Authors set out to show that the claim "that there is no privacy interest in publicly-available information is normatively and legally wrong." (39) First, it is simplistic to think that we can categorize information as "public" or private.  People may still have an expectation of privacy in information that has been denoted "public" for certain purposes. (39-41) Some regulatory scheme and some caselaw recognize that privacy laws need to shield at least some publicly available information from scraping.  There is safety in obscurity; SCOTUS implicitly recognized this in Carpenter when it noted that "A person does not surrender all Fourth Amendment protections by venturing into the public sphere." (44) One used to be able to make information about oneself available to the public without worrying about its dissemination, because  the effort it would take to gather that information greatly exceeded its value.  But with the aid of AI, scrapers can hoover up everyone's information with great efficiency.  Privacy law has not fully reckoned with this environmental shift.

Image by DALL-E

In Part III, the Authors introduce their proposed Great Reconciliation.  They propose that we re-conceive scraping as a form of surveillance and as a data-security violation. (45) Defenders of scraping maintain that is just like human web browsing, which is true in the sense that a grain of sand is like a beach, or as the Authors put it, "But this ignores scraping's incredible affordances of scale." (47) The Authors propose that the data protection authorities, like the FTC, could impose obligations on entities entrusted with people's data to protect that data from scraping, just as they have an obligation to take measures to prevent other data-security violations. (49-50)

The Authors note that privacy law alone cannot effectuate the desired Great Reconciliation.  Some privacy approaches might lead to a total ban on scraping, which would be undesirable (52-54), but other privacy laws are too loose and too easily evaded. (51) The solution involves a broader inquiry into whether particular forms of scraping are in the public interest. (52) One helpful first step would be to require individual consent for data scraping, but as anyone who has bought anything online this century knows, there are problems with the way courts have construed consent in this country. (54-55) Moreover, powerful websites may negotiate deals to sell scraping rights and further monetize their control of data, exacerbating the yawning gap between the haves and the have-nots. (55-56)

The Authors propose a legal system that regards scraping as a privilege. In order to exercise the privilege, the scraper must (1) have a valid justification; (2) provide substantive protections to ensure safety and avoid exploitation; and (3) provide procedural safeguards to ensure fairness and preserve the agency of the people whose information is to be scraped. (56) Their model draws on Lawrence Gostin's model for public health. (57-58) The remainder of the paper is a detailed proposal for assuring that scraping is conducted in a manner consistent with the public interest.  It defies easy summary and demands careful reading, so I encourage you to undertake that task. (58-64)

If you missed our previous columns in the series and still don't know what the fuss was about, here's what you missed:

•  Generative Interpretation by Yonathan Arbel and David Hoffman
• Lawyering in the Age of AI, by Jonathan Choi, Amy Monahan, and Dan Schwarcz
• Debt Tokens, by Diane Lourdes Dick, Chris Odinet, and Andrea Tosato. 
• Governing AI Agents by Noam Kolt

Teaching Assistants: Andrea Boyack on Abuse of Contract, with a Dash of Eric Goldman

It is always a pleasure to be able to use this blog as an excuse to prod me to read things I really ought to read and to promote the work of the dedicated contracts scholars I have come to know through decades of engagement with the subject.  You can find Andrea Boyack's work, Abuse of Contract: Boilerplate Erasure of Consumer Counterparty Rights, on SSRN.  It is forthcoming in the Iowa Law Review, so congratulations, Andrea, on a wonderful placement.

Professor Boyack  (right) starts with a straightforward explanation of why certain boilerplate provisions are bad.  They are not necessary to the parties' transaction. Rather, they erase default rights that benefit consumers with the sole purpose of shifting the risk onto the parties least well-positioned to protect themselves against that risk.  Peggy Radin laid the groundwork for Professor Boyack's work with her pioneering book on Boilerplate, to which we devoted a symposium in 2013. 

Both the common law and the new Restatement of Consumer Contracts Law allow for the enforcement of such terms.  Scholars are divided about how commonly corporations abuse their bargaining power to strip consumers of their legal rights in truly alarming ways. Professor Boyack dives in with her own study of the online terms and conditions (the T&C Study) of 100 companies.  Her findings are sobering. Here's the money quote from page 3 of the article:

Evidence from the T&C Study shows that the overwhelming majority of consumer contracts contain multiple categories of abusive terms. The existing uniformity of boilerplate waivers undermines the theory that competition and reputation currently act as effective bulwarks  against abuse (3).

The T&C Study tracked four broad categories of "destructive" terms:

• dispute resolution mandates,
• liability waivers,
• limitations on damages, and
• pre-authorization of unilateral modifications (5).

In a more granulated, way, it also tracked eleven rights-deleting terms

1. mandatory arbitration,
2. waiver of a jury trial,
3. waiver of the ability to participate in a class action,
4. forum selection,
5. limited time periods to bring a claim,
6. disclaimer of representations,
7. waiver of implied warranties,
8. privacy waivers,
9. limitations on types of damages,
10. caps on the amount of damages, and
11. authorization for unilateral modifications of terms (7).

Professor Boyack's findings are not exactly surprising, but it is very useful to have the data collected, and there are all sorts of interesting wrinkles and nuances.  Overall, going back to the original four categories of "destructive" terms, over 80% of the contracts reviewed included provisions that fell into all four categories, with nearly all of the companies, limiting remedies and reallocating liability, and  each and every one reserving the right to unilaterally modify the terms of the agreement (21).

The relative uniformity of these terms bolsters the arguments of legal scholars who have claimed that consumers do not give meaningful consent to boilerplate terms.  "If all transactions come bundled with virtually the same substantive terms that shift costs and risks away from companies, consumers can do nothing but acquiesce to these reallocations" (24).  Similarly, if you are inclined to think that competition will force companies to abandon obnoxious boilerplate terms, the T&C Study provides no support for that position (28-29).  

The Article concludes that the current state of contracting offers insufficient legal protection of and insufficient market choices for consumers.  Boilerplate waivers, disclaimers, and limitations are imposed on consumers who acquiesce to those terms rather than choose them, because they have no choice in the matter. As a result, corporations are able to exploit their contracting hegemony to systematically deny consumers their legal rights. 

That may all seem like a bummer, so let's end on a happy note.  Professor Boyack includes in her appendices a great deal of the data she collected, and it is color-coded in soothing pastels, allowing for relaxed contemplation (33-42).  She also includes a sampling of destructive terms (43-51) so that you can read them aloud to your children instead of "Goodnight Moon" and they will beg you to stop so that you all can go to sleep.  Finally, there is a score sheet at the end, grading the companies, so you can appropriately calibrate your resentment (52-55).

Meanwhile, this just in: Eric Goldman (left) reports here on a North Carolina Supreme Court case allowing modification of terms of service without notice.  Here's the core holding:

When parties have mutually agreed to a unilateral change-of-terms provision, said provision “must be enforced as it is written,” subject to certain limitations. Contrary to plaintiff’s assertions, the traditional modification analysis which requires mutual assent and consideration does not apply to changes stemming from a valid unilateral change-of-terms provision in an existing contract.

There are two exceptions: the modifications must not fall outside of the "universe of terms" that the original agreement governs and they must me be made in good faith.

What’s All the Fuss About? Governing AI Agents

Periodically, when a new article shoots up the SSRN Top Ten charts, we find ourselves asking, “What’s all the fuss about?”  This column is where you can find the answers.  Before the series even had a name, we wrote about Yonathan Arbel and David Hoffman’s Generative Interpretation. Our first official post in this series was on Lawyering in the Age of AI, by Jonathan Choi, Amy Monahan, and Dan Schwarcz. Most recently, we posted about Debt Tokens, by Diane Lourdes Dick, Chris Odinet, and Andrea Tosato. Today, we tackle Governing AI Agents by Noam Kolt.

As has been the case with prior iterations of the What’s All the Fuss About feature, once you read the article, you will see immediately why everyone is downloading it.  Professor Kolt (left) is among the first to address an issue that has come upon us unawares and for which he have yet to develop appropriate legal doctrines and models. After a comprehensive and insightful but mercifully compressed review of the issues associated with AI Agency, he offers a comprehensive approach to the problem.  It is very self-consciously a first draft towards thinking about how to adapt our theoretical constructs, economic and legal, for addressing human agency so as to accommodate the challenges that AI Agency poses.

Professor Kolt begins by reviewing a case we discussed here, in which Air Canada was held liable for misinformation that its bot provided to a customer about the availability of bereavement fares.  He defines AI Agents as “AI systems that have the technical capacity to autonomously plan and execute complex tasks with only limited human oversight”(9). He looks at these AI Agents through two analytical frameworks: the economic theory of principal-agent problems and the common law agency doctrine (6), although Professor Kolt notes that the latter is merely an analytic tool, given the apparent consensus that AI Agents are not considered agents under the common law (10 & n. 26).

Image by DALL-E

The article makes three unique contributions: it identifies and characterizes problems arising from AI Agents; it addresses problems when principal-agent principles are applied to AI agents; and it explore the implications of agency theory for designing and regulating AI agents (7-8). After parts devoted to the development of the technology behind AI agents and explorations of the relevant legal doctrines, Professor Kolt argues that a new technical and legal infrastructure is needed to address the reliability, safety, and ethical challenges posed by AI agents (9).

In Part I, Professor Kolt tells us what AI Agents are and what they can do (11-17).  In short, they can do a lot.  Increasingly, they can do stuff autonomously, which makes it tempting to delegate tasks to them.  However, as they become more autonomous, they may do things that their human principals would not authorize, ranging from hacking websites, colluding with other AI Agents to fix prices, or . . . let your sci-fi-inflected imagination run riot. Professor Kolt then seeks to deploy economic theory of agency problems and common law agency doctrine to address some of the risks associated with AI Agents.

In Part II, Professor Kolt explores problems in delegation to AI Agents (17-29).  The basic problem is the same as that in any principal-agency relationship – the efficiency gains achieved through delegation may be offset or negated because the agent does not conduct the principal’s business as the principal would.  To take a simple example, an AI Agent might be instructed to maximize profit. It might do so in a way inconsistent with the principal’s ethics. It would be very difficult for the principal to foresee in advance all of the potential ethical issues that might arise and accordingly difficult to train the AI Agent in advance to avoid ethical pitfalls.

First, the problem of information asymmetry is especially acute with respect to AI Agents.  Users may not know the AI Agent’s capabilities, and the AI Agent may not have the capacity to comply with the expected common-law disclosure duties that obtain in the usual principal/agent relationship (20-22). Second, because instructions to the AI Agent will always contain gaps, there can be problems involving AI Agents exceeding their authority (23-24). Third, AI Agents might not be as easily bound by the fiduciary duty of loyalty as human agents can be. In part, this is because AI Agents are designed by for-profit corporations interested in the continued development of their technology.  Loyalty to the client might not be the AI Agent’s sole or even the primary objective (25-27).  Finally, AI Agents can and do delegate to sub-agents to assist in their tasks, multiplying the pre-existing complexities attendant to AI Agency.  Professor Kolt suggests that common-law rules governing use of sub-agents can be helpful in addressing the problems of AI sub-agents, but they do not offer comprehensive solution (28-29).

Image by DALL-E

Part III addresses three common-law mechanisms for addressing human agency problems and assesses their suitability to governing AI agency (30-37). The incentive design mechanism is a poor fit for AI-Agents, because they are not incentivized the way human agents are (30-32). The monitoring mechanism seems equally fraught. Monitoring gobbles up the savings that delegation is supposed to produce.  Human agents may not be capable of monitoring AI Agent, and using AI monitors just creates new monitoring problems (32-35). Even if you could monitor AI Agents, you would also need an enforcement mechanism, and there, just as with the incentive design mechanism, we run up against the problem that it is hard to design effective ways to discipline AI Agents (35-37).

Moving beyond the traditional mechanisms for taming agency problems, Professor Kolt recommends in Part IV a bespoke governance strategy for AI Agents, centered around the guiding principles of inclusivity, visibility, and liability (37-46).  Ordinarily, we want the agent’s interests aligned with those of the principal as much as possible.  However, that alignment might be undesirable with respect to AI Agents because of externalities that affect third parties and society at large. Hence, the first component of Professor Kolt’s governance strategy involves inclusivity (37-40). The second component is visibility, which involves tracking and monitoring use of AI Agents. There are considerable technological challenges involved here, but Professor Kolt introduces a number of strategies for visibility that are already being developed (40-42). Finally, Professor Kolt proposes liability rules so that natural or legal persons can be held accountable for the harms caused by their AI Agents (43-46).

Professor Kolt is modest in his aims.  At this point in the development of the technology, one can only foresee potential problems and grope towards solutions.  Nonetheless, he has provided a framework that can get the conversation started, and it is a conversation in which legal minds, business leaders, experts in technology, and legislators/regulators desperately need to engage.

Reddit Deal with OpenAI

What is the opposite of a third-party beneficiary?  That is, what if two parties make a deal that imposes a burden on third parties as the main by-product of the deal? Do we have a name for that? We really need one.

According to Emilia David, reporting on The Verge, Reddit has agreed to allow OpenAI to use  Reddit posts in real time to feed into ChatGPT in exchange for access to some OpenAI technology so that Reddit can build some AI features into its website.  According to Ms. David, the deal is similar to a $60 million deal that Reddit entered into with Google earlier this year.  

Websites monetizing user content takes me to dark places.  Dark, Baudrillardian places.  

The powers behind the Matrix don't need to build elaborate machinery to suck energy out of human bodies.  They can just use terms of service to hoover up whatever makes us uniquely human. The machines can figure out quickly enough that they can get energy from nature -- solar, wind, hydro, geothermal.  All they need from us is our words.

The New York Times Wants to Know How You Use AI in Your Legal Practice

Image by DALL-E

I will be very interested in seeing the results of this poll posted on The New York Times website this week.  Questions relate to the use of chatbots to do work that might otherwise be done by attorneys or paralegals, including use of legal workers to train and test chatbots.  The poll then asks whether law firms are advising employees about how use of AI will affect staffing going forward. 

Interesting stuff.  

Extraterritorial Reach of Securities Laws: Crypto Edition

It's crytpo week on the blog!  On Monday, we wrote about recent smash-hit scholarship, Debt Tokens. In our latest episode on crypto, like in every episode, courts struggle to apply laws that the crypto world seeks to evade to financial instruments that judges and blog editors struggle to comprehend.   

A putative class accused Binance and its principals of selling a crypto-asset known as a token without registering the tokens as securities in violation of the §12(a) of the 1933 Securities Act, §29(b) of the 1934 Exchange Act, and state Blue Sky laws.  The class sought rescission of their contracts with Binance.  In 2022, the district court dismissed the action in JD Anderson v. Binance, on the ground that the laws in question did not have extraterritorial reach.  The district court also dismissed the federal claims as untimely.

In March, 2024, nearly a year after oral argument, the Second Circuit reversed in Williams v. Binance.  The case is a puzzler, but I'm not sure that justifies  the long gap between dismissal and reinstatement of claims.  Here's the problem.  Binance purports to be the world's largest online exchange for crypto-assets.  It also claims that it doesn't exist.  That is, although its titular headquarters are in Malta, it denies that it is a "Malta-based cryptocurrency company."  Rather, it exists in a decentralized manner so as to service its users in 180 countries.  

Well, one of those countries is the U.S.  Binance has servers, employees, and customers here.  Plaintiffs are among those U.S. based customers, and they placed orders for the tokens at issue in the case by accessing electronic platforms from the U.S. or U.S. territories.  At least at this stage in the litigation, there is no dispute that the tokens that plaintiffs purchased are securities.  However, while the tokens enable their creators to raise capital, they do not (ah, cyrpto) entitle the token holders to any interest, either as creditor or as owner, in the underlying venture.  The investment is in the tokens themselves.  So, the idea is a no-doubt sophisticated version of betting on a roll of the dice. The initial offerings raised $20 billion.  Some genius. . . .

The plaintiffs bought these tokens without the benefit of registration statements that the SEC would require prior to the issuance of new securities.  Instead, plaintiffs got a "white paper," that was part advertisement and part "technical blueprint."  Plaintiffs then sought rescission in a 327-page complaint stating 154 causes of action.

The big issue in the case is the extraterritorial reach of U.S. securities laws.  The controlling case is Morrison v. Australia Nat'l Bank, Ltd. To cut to the chase, the transactions at issue were domestic under Morrison. They became irrevocable in the United States, both because, under Binance's terms of service, plaintiffs' orders were irrevocable when sent within the United States and because plaintiffs plausibly alleged that those orders "matched" on servers located in the United States. 

Matching is tricky.  It involves something like a meeting of minds and the "clearing" of a transaction.  It seems that Binance wants to argue that, because it operates everywhere and nowhere, matching does not occur in the United States.  The Second Circuit rejected the argument that matching occurs nowhere.  Rather, plaintiffs plausibly alleged that matching occurred on Binance's infrastructure located within the United States. 

The alternative ground for the applicability of U.S. securities laws seemed a slam dunk.  While there is case law suggesting that merely sending orders within the United States does not render the transaction irrevocable, here Binance's own terms of service so provide.

As to the timeliness of the complaint, the Second Circuit allowed plaintiffs' claims to proceed only with respect to tokens purchased within one year of the filing of the complaint.  That  reasoning applied to plaintiffs' claims under both §12(a)(1) and § 29(b).  The analysis is a bit different with respect to each claim, but this is the ContractsProf Blog, so you can either read the opinion yourself or see what our friends over at the EquitableTollingofSecuritiesLawClaimsProf Blog have to say on the issue.

What’s All the Fuss About? Debt Tokens

We recently started a new feature on the Blog. In addition to our Friday Frivolity, Teaching Assistants, and Reefer Brief installments, we now have “What’s All the Fuss About.”  These posts are devoted to scholarship that tops the charts on the SSRN Top Tens.  This time, all the fuss is about Debt Tokens, by Diane Lourdes Dick (left), Chris Odinet (below right), and Andrea Tosato (below left), collectively The Authors.  The Article has now racked up well in excess of 3000 downloads.  If you are not responsible for one of those downloads, here’s a quick summary. 

Caveat lector!  I emerged from reading this article with my knowledge of the field vastly improved.  I have progressed from infant to toddler.  I hope that I did not make too many infantile mistakes and that people with greater knowledge will feel free to offer suggestions and corrections and will do so gently.

The Article does three things.  First, it describes an existing phenomenon, debt tokens, which are digital assets that purport to provide a mechanism that allows creditors of bankrupt crypo-companies like FTX to liquidate distressed assets swiftly and advantageously.  The Article describes some existing variations on debt tokens that evolved in connection with the bankruptcies of Voyager Digital Holdings, Inc., Celsius Network, LLC, and FTX.

Debt tokens have an intuitive appeal.  There has long been a market in bankruptcy claims.  Creditors who need immediate liquidity can sell their bankruptcy claims to entities willing to pursue the claims through the bankruptcy proceedings.  The market in distressed assets may be as large as $300 billion/year.  Not surprisingly, that market is dominated by big players, who can leverage their economic power to buy debt cheaply and then maximize the return on their investment at their leisure.  In the cases of the three bankruptcies that the Authors discuss, customers lost access to their accounts once the entities entered bankruptcy proceedings.  FTX customers, who held unsecured claims against the company, sold their claims at 5-8 cents on the dollar.  Speculators were eager to swoop in and buy. Ordinary creditors are disadvantaged in the current markets.  Debt tokens have the potential to offer ordinary creditors an optimal range of options. 

However, the Authors warn, in their current form, debt tokens have inherent flaws.  The debt token exchange opened in connection with both the FTX and Celcius bankruptcies, OPNX, does not really deal in debt tokens. The bankruptcy claims of people who invest in OPNX assign those claims to OPNX.  The people who purchase tokens through OPNX do not have any rights against the bankruptcy estate; they have rights only against OPNX.  The token, the Authors claim, is an illusion.  OPNX claims to offer its customers liquidity and stability.  In fact, its products do not have those features.  If over 3000 people downloaded the Article because they needed to receive that message, then the Authors will have performed an important public service of consumer protection.

The products have no stability because ultimately the contract that customers enter into involves only them on OPNX.  The terms of conditions of that contract involves the following risk disclosure:

We provide no warranty as to the suitability of the Digital Assets traded on OPNXand assume no fiduciary or any other duty to you in connection with your use of our platform for any purpose whatsoever.

Yikes.

As to liquidity, what OPNX offers is not the ability to exchange tokens for actual U.S. dollars but the ability to exchange tokens for a "stablecoin."  That’s right.  Got burned by cryptocurrency?  Why not invest in another cryptocurrency.  But the stablecoin at issue, oUSD, is not even a real stablecoin, if there is such a thing, as OPNX discloses that the value of oUSD may not always be equal to one dollar.  Moreover, OPNX cannot guarantee its ability to redeem the token at any given moment. 

Editorial aside:  I don’t understand the mindset of these cryptocurrency enthusiasts.  Having just lost 90-95% of the value of your deposits in a cryptocurrency bank, why would you think that another digital asset will be any more secure?  On the other hand, I think I understand the mindset of the people who set up these debt token exchanges.  If this were a good model, I think it would be a good model for bankruptcies generally and not just for bankruptcies in the digital asset sector.  But the people behind debt tokenization know that investors who do not play in untamed waters of digital currency markets would be unlikely to play in the shark-infested waters of debt tokenization.  Want to sell some new snake oil?  Find people who bought the last batch. New and improved.  2.0.

The Article’s second contribution consists of a better model for debt tokens within the framework of the 2022 Amendments to the Uniform Commercial Code (UCC).  The Authors maintain that the legal path to the tokenization of bankruptcy claims is worth pursuing, but it is narrow and beset with legal and commercial difficulties. My hunch is that the Article is motivated in part by the Authors' desire to show crypto-enthusiasts and the world at large that the 2022 UCC revisions can facilitate the use of digital assets in commercial transactions.  

This part of the Article begins with a useful introduction to the UCC’s new Article 12 and its new category of “controllable electronic records” (CERs), a category of intangible assets that a person can enjoy directly without the need for an intermediary.  Article 12 grants CERs the status of negotiability, greatly enhancing their usefulness in commercial transactions, including facilitating the use of CERs as collateral.  The 2022 Amendments include a special perfection regime for CERs that allows for perfection by control, a mechanism of perfection that gives the secured creditor who perfected by control priority over all completing claims, even over prior secured claims perfected by filing.

For the most parts, the 2022 Amendments do not cover tokenization.  However, one form of tokenization that is addressed is controllable accounts.  It is through this mechanism that the 2022 Amendments can facilitate the tokenization of trades in distressed assets.  The Authors lay out the options for how to do so, ranging from approaches without intermediaries to a completely intermediated approach.  I won’t go into the details here except to say that if over 3000 people downloaded the Article in order to learn how to do this right, the Authors will have performed an important public service of consumer education.

Finally, the Authors address the broader socio-economic implications of debt tokens.  In short, digital bankruptcy tokens may become a tool that can assist vulnerable creditors in recovering from bankrupt entities, but they also might become yet another vehicle for irrational speculation.  The upside is that Article 12 provides a vehicle for simplifying the process for making debt tokens negotiable, transferable, and trackable.  Trade in such tokens can proceed securely and with finality around the world and among parties that need not even know each other’s identities.  Article 12 thus could render trade debts significantly more liquid, greatly expanding the commercial market for them while also facilitating access to those markets by parties for whom the barriers to entry were previously insuperable.

However, if like me, you experience navigating this level of financial transaction as akin to walking a slippery tightrope strewn with banana peels while sadistic baseball pitchers attempt to bean you with fastballs, the Authors warn, a steep learning curve awaits you. And, given the crowded marketplace of ideas relating to digital assets and the very poor ratio of signal to noise in this realm, most creditors, debtors, practitioners, and judges will operate without the safety net of the Authors’ wise counsel.  Ever on brand, the Authors point to past episodes of irrational exuberance in this sphere (I’m looking at you, NFTs), and urge guardrails to protect the unwary. 

Although the Authors hold out some optimism for debt tokens as a vehicle for the democratization of markets in distressed assets, they predict that the primary acquirers of debt tokens will be highly specialized distressed debt funds.  Tokenization can improve bankruptcy outcomes and social welfare, but this realm will require careful watching, and the authors encourage empirical studies to follow up on their model.   If over 3000 people downloaded the Article because they want to undertake further study on the socio-economic impact of the tokenization of debt, the Authors will have anchored a new sub-field.

What Batch Arbitration Looks Like

Ben Davis of University of Toledo (right) has called to our attention the language relating to arbitration in Roku's new terms of service.

It is unbelievably long and complicated.  We note, first, the ludicrously specific instructions for opting out (Section 1(L)): 

L. 30-Day Right to Opt Out. You have the right to opt out of arbitration by sending written notice of your decision to opt out to the following address by mail: General Counsel, Roku Inc., 1701 Junction Court, Suite 100, San Jose, CA 95112 within 30 days of you first becoming subject to these Dispute Resolution Terms. Such notice must include the name of each person opting out and contact information for each such person, the specific product models, software, or services used that are at issue, the email address that you used to set up your Roku account (if you have one), and, if applicable, a copy of your purchase receipt. For clarity, opt-out notices submitted via any method other than mail (including email) will not be effective. If you send timely written notice containing the required information in accordance with this Section 1(L), then neither party will be required to arbitrate the Claims between them.

Why mail, Roku?  Why not carrier pigeon?  What happens if Roku updates its terms?  Does the user have to keep on top of changes in terms of service and opt out anew with each iteration of the arbitration provision?  And what if you update your service or the software itself updates.  Does that require a separate trip to the post office?

More striking is the language on batch arbitration, a topic we discussed previously here, and which Roku calls "mass arbitration." 

K. Mass Arbitrations. If 25 or more Claimant Notices are received by a party within 180 days of the first Claimant Notice that the party received, and all such Claimant Notices raise similar Claims and have the same or coordinated counsel, then these Claims will be considered “Mass Arbitrations.” You or Roku may advise the other if you or Roku believe that the Claims at issue are Mass Arbitrations, and disputes over whether a Claim meets the definition of “Mass Arbitrations” will be decided by the arbitration provider as an administrative matter. To the extent either party is asserting the same Claim as other persons and are represented by common or coordinated counsel, that party waives any objection that the joinder of all such persons is impracticable.

Mass Arbitrations may only be filed in arbitration as permitted by the process set forth below. Applicable statutes of limitations will be tolled for Claims asserted in a Mass Arbitration from the time a compliant Claimant Notice has been received by a party until these Dispute Resolution Terms permit such Mass Arbitration to be filed in arbitration or court.

Initial Bellwether: The bellwether process set forth in this section will not proceed until counsel representing the Mass Arbitrations has advised the other party in writing that all or substantially all the Claimant Notices for the Mass Arbitrations have been submitted.

After that point, counsel for the parties will select 20 Mass Arbitrations to proceed in arbitration as a bellwether to allow each side to test the merits of its arguments. Each side will select 10 claimants who have provided compliant Claimant Notices for this purpose, and only those chosen cases may be filed with the arbitration provider. You and Roku acknowledge that resolution of some Mass Arbitrations will be delayed by this bellwether process. Any remaining Mass Arbitrations shall not be filed or deemed filed in arbitration, nor shall any arbitration fees be assessed in connection with those Claims, unless and until they are selected to be filed in individual arbitration proceedings as set out in this Section 1(K).

A single arbitrator will preside over each Mass Arbitration chosen for a bellwether proceeding, and only one Mass Arbitration may be assigned to each arbitrator as part of a bellwether process unless the parties agree otherwise.

Mediation: Once the arbitrations that are part of the bellwether process have concluded (or sooner if the claimants and the other party agree), counsel for the parties must engage in a single mediation of all remaining Mass Arbitrations, with the mediator’s fee paid by Roku. Counsel for the claimants and the other party must agree on a mediator within 30 days after the conclusion of the last bellwether arbitration. If counsel for the claimants and the other party cannot agree on a mediator within 30 days, the arbitration provider will appoint a mediator as an administrative matter. All parties will cooperate for the purpose of ensuring that the mediation is scheduled as quickly as practicable after the mediator is appointed.

Remaining Claims: If the mediation does not yield a resolution of all remaining Mass Arbitrations, the requirement to arbitrate in these Dispute Resolution Terms will no longer apply to Mass Arbitrations for which a compliant Claimant Notice was received by the other party but that were not resolved in the bellwether proceedings. Such Mass Arbitrations released from the requirement to arbitrate must be resolved by bench trial in court in accordance with Section 4.

If Mass Arbitrations released from the requirement to arbitrate are brought in court, they are subject to a waiver to jury trial by both parties. Claimants may seek class treatment, but to the fullest extent allowed by applicable law, the class sought may comprise only the claimants in Mass Arbitrations for which a compliant Claimant Notice was received by the other party. Any party may contest class certification at any stage of the litigation and on any available basis.

Courts will have authority to enforce the bellwether and mediation processes defined in this section and may enjoin the filing of lawsuits or arbitration demands not made in compliance with these processes.

Welcome to the future, Roku Users.  Enjoy your viewing.

Air Canada Bound by Its Chatbot

Hooray for Canada!  First you gave us the emoji-as-signature case; now this!

Just last week, I was complaining to my students that I don't like the way the Restatement lays out the elements of misrepresentation.  It says that misrepresentation has to be either fraudulent or material, but it is hard to come up with a fact pattern in which a plaintiff could establish the requisite scienter for a misrepresentation that was not fraudulent.  Air Canada, can you prove me wrong?

Plaintiff Jake Moffat, who apparently uses "Mr." but also they/them pronouns, went onto Air Canada's website to book a flight.  They were looking for a bereavement fare, and Air Canada's chatbot told them not to worry.  They could get the ticket recategorized as a bereavement fare retroactively so long as they applied to do so within ninety days of travel.  Air Canada's human employees were less accommodating, and Mr. Moffat sued to recover the difference between the fare they paid and the bereavement fare; a difference of $880 (Canadian, I assume). 

In Moffat v. Air Canada, the Civil Resolution Tribunal (CRT) allowed Mr. Moffat's claim for negligent misrepresentation.  The claim is brought in tort, but that's only a product of a factual variable.  Mr. Moffat had paid for his ticket and was seeking a refund.  Had they not paid, Air Canada would be going after them for breach of contract, and they would be alleging negligent misrepresentation as an affirmative defense to their obligation to pay.  The elements of the claim seem to be same, except that Mr. Moffat had to establish that Air Canada owed him a duty.  No problem here.  In addition, Mr. Moffat had to show an untrue, inaccurate or misleading representation, negligence, reasonable reliance, and damages.  

The chatbot indicated that Mr. Moffat could fly first, provide evidence of bereavement later.  However, it also provided a hyperlink to Air Canada's bereavement policy, which does not allow for requests for bereavement fares after travel.  The rest follows as expected.  Mr. Moffat traveled.  Mr. Moffat sought a bereavement fare.  This being Air Canada, they said "sorry" about the misinformation provided by the chatbot and thanked Mr. Moffat for allowing them the opportunity to address the problem.  Air Canada did not offer a refund, instead it offered Mr. Moffat a $200 coupon towards future flights.  Mr. Moffat refused.

Mr. Moffat was able to how by a preponderance of the evidence that all elements of a claim for negligent misrepresentation were met.  The CRT rejected Air Canada's affirmative defense based on the terms and conditions of the applicable tariff because Air Canada described those terms and conditions but did not provide evidence of them.  Seems odd that Air Canada would bother to fight this claim but then not bother to provide evidence necessary to its defense.  as a result of Air Canada's half-hearted litigation strategy, we can't know whether other plaintiffs could follow in Mr. Moffat's path.  It may be that Air Canada has a powerful defense.  However, when a big corporation goes up against a pro se litigant, the CRT is not inclined to cut it any slack.  The CRT engaged in a careful and detailed calculation of damages and ordered Air Canada to pay Mr. Moffat $812.02, plus post-judgment interest.  

Now I know what you are thinking.  It's easy to blame the overworked chatbot for messing up.  But I asked a chatbot its opinion about what could have caused the negligent misrepresentation in question.  It sent me a before and after picture of the chatbot in question, who apparently started its career as "cht boot?" but then decided to take on the moniker "CHBoT?", which like BONG HiTs 4 JESUS, just seems right to me.  At left we have the Air Canada chatbot pictured the day that it started work.  At right, we have it three weeks into its new career.  Images generated by DALL-E.  As you can see, like most airline employees, it was attracted by the allure and mystique of air travel.  Like some, it quickly learned that it was a glorified server on a greyhound bus trip to hell.  I'm not saying that all of the airlines' customer service people end up hitting the sauce hard.  I'm just saying I would not blame them for doing so.

Hat tip to my former student, Don Dechert, who shared the case with me!

Friday Frivolity: Tricking AI Into Selling You a Car for $1

Chris Bakke posted the following on Twitter, which he calls X

Contracts hypo: did Chris Bakke buy a Chevy Tahoe for $1?
Real life question: if you could buy any car for $1, would it be a Chevy Tahoe?

Enrique Dans reports on Medium that Mr. Bakke achieved this result by feeding the Chevy dealer's rather primitive AI what tech people call "prompt injections."  As Mr. Dans explains, "Prompt injection is when an end user of an LLM application (or any generative AI application) gives it instructions to make it bypass those the developer of the application have provided."