Transatlantic Antitrust and IPR Developments, Newsletter Issue No. 2/2023 (December 19, 2023)

Contributors:

Alexandros Kazimirov,
Amedeo Rizzo, Irene Kamara,
Marie-Andrée Weiss,
Zihao Li

Editor-in-chief: Juha Vesala

Contents     

Antitrust

United States

Athletes, Ivies and the NCAA

United States v. Google: Predictions Before the Showdown

Franchise Agreements: The Case for Limited Non-Compete Clauses

FTC & DOJ Review of Merger Guidelines 2023

Law School Boycotts and the Sherman Act

Intellectual Property

United States

AI, Face Swapping, and Right of Publicity

Other Developments

United States

SEC’s First Enforcement Case under Regulation Best Interest

European Union

Cyberstalking and Online Platforms’ Due Diligence in the EU Digital Services Act

EU Adoption of DAC 8 – Mandatory Exchange of Information between Tax Authorities on Crypto Assets

Large Language Models and the EU AI Act: the Risks from Stochastic Parrots and Hallucination

About the contributors

Alexandros Kazimirov is an attorney admitted to practice in California. He studied civil law in Cyprus and Athens (Greece), has received an LL.M. from Berkeley Law School with a concentration in Securities Regulation and a Certificate in Regulatory & Antitrust Law from Cornell Law School. During his studies, he clerked at the Court of Justice of the European Union. He serves on the Intellectual Property Committee of the Antitrust Section of the American Bar Association and the Securities Litigation Section of the San Francisco Bar Association. His research examines the approach taken by the Securities and Exchange Commission to regulate digital asset markets compared to the European Union’s Markets in Crypto-Assets Regulation. He has been a TTLF Fellow since September 2023.

Amedeo Rizzo is a D.Phil. in Law and Academic Tutor at the University of Oxford, UK, where he conducts research in taxation, innovation, and development. He is an Academic Fellow of Taxation at Bocconi University, Italy, and SDA Fellow of Tax and Accounting at SDA Bocconi School of Management, where he coordinates the Accounting & Tax Policy Observatory and the Transfer Pricing Forum. He is the director of the Innovation Policy Network and a member of the Group of Experts on anti-corruption for Transparency International Italy. As a TTLF Fellow, his research focuses on the analysis of different types of tax incentives to enhance innovation through intellectual property and research and development activities. Previously, he worked for the Directorate-General for Economic and Financial Affairs (DG ECFIN) of the European Commission as an external advisor on budget and tax policy, and for the International Tax and Transfer Pricing Team of PricewaterhouseCoopers, Milan. He also collaborated with the Centre for Budget and Governance Accountability, India, on financial transparency issues in Asia-Pacific countries. Prior to his D.Phil. in Law at the University of Oxford, Amedeo obtained an M.Sc. in Taxation from the University of Oxford (distinction), an M.Sc. in Business Administration and Law (summa cum laude) and a B.Sc. in Business Administration, both from Bocconi University.

Irene Kamara (Dr.) is Assistant Professor at the Tilburg Institute for Law, Technology, and Society in The Netherlands. Her research focuses on the protection of human rights in the digital environment, and substantive and procedural aspects of cybercrime. Irene is teaching cybercrime and cybersecurity law, legal aspects of technical standardization, and international personal data flows at master and bachelor programs. Irene has previously worked as legal researcher at the Law and Criminology Department of the Vrije Universiteit Brussel in Belgium and as attorney-at-law in law firms in Athens, Greece. She has conducted research as principal investigator for the European Cybersecurity Agency ENISA, the European Commission, the National Cyber Security Agency (NCSC) and the National Coordinator for Counterterrorism and Security (NCTV) in The Netherlands. She holds a joint PhD in law from the Vrije Universiteit Brussel and the University of Tilburg (2021), a LL.M. in Law and Technology from the University of Tilburg (2014), a MSc in International and European Studies from the University of Piraeus in Greece (2013), and a Bachelor of Laws from the Demokritus University of Thrace in Greece. In 2021, CEN and CENELEC honoured Irene with the prestigious Standards + Innovation award for the category Individual Researcher Innovator, the first legal scholar to receive this award. Irene has been a TTLF Fellow since 2022.

Marie-Andrée Weiss is an attorney admitted in New York and in Strasbourg, France. Before becoming an attorney, she worked for several years in the fashion and cosmetics industry in New York as a buyer and a director of sales and marketing. She graduated from the University of Strasbourg in France with an M.A. in Art History, a J.D. in Business Law, an LL.M. in Criminal Law, and an LL.M. in Multimedia Law. Marie-Andrée also graduated from the Benjamin N. Cardozo School of Law in New York City with an LL.M. in Intellectual Property Law. She is an attorney in New York and her solo practice focuses on intellectual property, privacy, data protection, and social media law.Zihao Li is a Lecturer (Assistant Professor) in Law and Technology at CREATe Centre, School of Law, University of Glasgow, UK. He has been a TTLF Fellow at Stanford Law School since 2023. Meanwhile, he has been invited as a Guest Lecturer to Trinity College Dublin, Ireland. Zihao is qualified in both Computer Science and Law. With his interdisciplinary background, his research interests concentrate on the intersection of law, data and information technology. Recently, his research mainly includes data protection law, AI and regulation, algorithmic pricing, Internet and intellectual property, and blockchain and law. Zihao’s research has been published in top-tier interdisciplinary academic journals, conference proceedings and books, including Nature Machine Intelligence, Computer Law and Security Review, IEEE Communications Magazine, IEEE International Conference on Communications, and European Data Protection Law Review. His research has been cited around the world. He has also spoken at several prestigious universities, including the University of Cambridge, UK, and King’s College London (KCL), UK. Zihao has been awarded the Modern Law Review (MLR) Scholarship operated by the LSE. Additionally, Zihao is a co-founder of the Scottish Law and Innovation Network (SCOTLIN), and the founding president of the Intellectual Property Society at the University of Glasgow. At Stanford Law School, Zihao principally investigates data privacy issues in algorithmic pricing within the EU and US.

Antitrust

United States

Athletes, Ivies and the NCAA

By Alexandros Kazimirov

On June 29, 2023, a student class action was filed against educational institutions that form the Ivy League, asserting that their agreement to restrict athletic scholarships for student athletes competing in Division I of the National Collegiate Athletic Association (NCAA) is a form of price-fixing, incompatible with antitrust law, under a per se and rule of reason analysis. The complaint echoes criticism of what is claimed as “procompetitive justification” for imposing limits on education-related expenses in National Collegiate Athletic Association v. Alston.

The question of compensation of student athletes has long permeated public discourse and the U.S. courts. At its core it is a matter of balancing between the interests of student athletes and the concept of amateurism, which constitutes intercollegiate sports under the NCAA. Some hold that refusing compensation to student athletes when the NCAA’s product is worth billions, is preposterous. The view that there is great disparity between the NCAA’s earnings and the student athletes’ compensation is one shared by Justice Kavanaugh, who laid out his separate opinion in Alston: “Those enormous sums of money flow to seemingly everyone except the student athletes. College presidents, athletic directors, coaches, conference commissioners, and NCAA executives take in six- and seven-figure salaries. Colleges build lavish new facilities. But the student athletes who generate the revenues end up with little or nothing”.

The NCAA on its part has historically focused on the premise of amateurism. According to the NCAA, amateurism defines college athletics, and the sport derives its value from it. Subsequently, the NCAA has been at arms with every court decision limiting NCAA’s discretion on how to address the issue.

A fundamental principle in all relevant cases so far, is that the courts discern between education-related expenses and education-unrelated expenses, such as endorsement deals based on a student-athlete’s name, image or likeness (NIL). What is more, education-related expenses can be in the form of tuition discounts or athletic scholarships.

The issue of whether college athletes should receive additional compensation beyond tuition, room and board, and other educational expenses has not been fully settled. In O’Bannon v. NCAA [802 F.3d 1049 (9th Cir. 2015)] the Ninth Circuit Court of Appeals lifted caps on education-related compensation, but allowed some discretion to the NCAA on limiting students’ earnings based on NIL deals which are unrelated to their education expenses. In Alston, the NCAA appealed the loss of their ability to cap educational-related expenses. And lost.

The Court described that just because the NCAA’s arrangement was not found to be illegal per se in a previous case (NCAA v. Board of Regents of the University of Oklahoma) this does not preclude the Court from examining it under the rule of reason.

And under the rule of reason, the Court laid out the process that the District Court employed to reach its conclusions. First, the students argued that the restraints on education-related expenses and additional compensation were anti-competitive measures by a party enjoying monopsony. Then, letting the NCAA rebut the argument by reiterating the pro-competitive aspects of the arrangement, such as retaining “amateurism”, i.e. the collegiate non-professional nature of sports. Third, carefully considering whether the consumer market can be attained using substantially less restrictive means.

The Court found in Alston that the District Court struck a good balance by condemning the restraints on educational-related expenses, but still leaving some discretion to the NCAA on how to manage such expenses, without having judges second-guess every decision they take.

Justice Kavanaugh’s concurring opinion is worthy of mention. He reminded the Court that NCAA’s appeal pertained only to caps on cost of attendance, which the Court unanimously rejected. But he went beyond this, saying he didn’t find the NCAA’s procompetitive justification for its rules limiting athletic scholarships or even NIL compensation persuasive either. In a nod to a future class of petitioners, Justice Kavanaugh indicated that there is more to be gained. His remarks were noticed and his call was answered in Choh v. Brown University.

From here on, with O’Bannon and Alston as precedent, it will be a tough road for the defendants to keep restrictions on athletic scholarships or NIL deals.

United States v. Google: Predictions Before the Showdown

By Alexandros Kazimirov

On September 12, 2023 the U.S. District Court for the District of Columbia heard the government’s case against Google, three years after it was filed. Like U.S. v. Microsoft in its time, this case may be pivotal for antitrust enforcement of Big Tech.

Rules

The Supreme Court defines monopoly power as “the power to control prices or exclude competition.” More precisely, a firm is a monopolist if it can profitably raise prices substantially above the competitive level. The offense of monopolization has two elements: (1) the possession of monopoly power in the relevant market and (2) the willful acquisition or maintenance of that power by means other than a superior product, business acumen, or historic accident.

Whether any particular act of a monopolist is exclusionary, rather than merely a form of vigorous competition, can be difficult to discern. As the court held in Microsoft:

First, to be condemned as exclusionary, a monopolist’s act must have an “anticompetitive effect.” That is, it must harm the competitive process and thereby harm consumers;

Second, the plaintiff, on whom the burden of proof of course rests, must demonstrate that the monopolist’s conduct indeed has the requisite anticompetitive effect […] it must demonstrate that the monopolist’s conduct harmed competition, not just a competitor;

Third, if a plaintiff successfully establishes a prima facie case under §2 by demonstrating anticompetitive effect, then the monopolist may proffer a “procompetitive justification” for its conduct. If the monopolist asserts a procompetitive justification (a non pretextual claim that its conduct is indeed a form of competition on the merits because it involves, for example, greater efficiency or enhanced consumer appeal) then the burden shifts back to the plaintiff to rebut that claim;

Fourth, if the monopolist’s procompetitive justification stands unrebutted, then the plaintiff must demonstrate that the anticompetitive harm of the conduct outweighs the procompetitive benefit.

Analysis

The DOJ asserts that Google has violated Section 2 of the Sherman Act by being a monopoly and it invokes three dimensions where Google’s harmful effect may be ascertained. Advertisers, consumers and competitors (other search companies). The DOJ claims that Google has harmed them through two markets: general search services and search advertising services. The DOJ finally states that Google has employed anti-competitive means like exclusionary agreements and tie-ins with phone manufacturers to block rivals from competition.

Courts have generally identified these actions to be problematic: refusal to deal; ceasing existing cooperation with competitors; predatory pricing; bundle discounts; tie-in arrangements and exclusive agreements. Tie-in arrangements are unlawful per se under three conditions: (1) separate products A and B, (2) market power in product A, and (3) result of the tie-in is to foreclose “a not insubstantial dollar volume of commerce in B’s market.” For exclusive agreements, a court will consider: (1) percentage of market foreclosed by these arrangements, (2) long term contracts, (3) feasibility of contracting with a competitor (4) alternatives or disintermediation.

Each dimension or class requires a separate analysis under the relevant market definition, and the respective conduct in question. Namely, the government must show how and to what extent Google’s conduct has harmed each class. And Google must have no good defense to come up with. If it does, the government must rebut the defense on proportionality grounds.

With respect to harm caused to consumers’ privacy (i.e. google search users) Google’s defense will be that (i) users do not pay to use the search engine, (ii) despite the fact that DuckDuckGo once found that it took 15 taps to switch the default search engine on Android, ultimately users still retain the choice to use their search engine of preference and therefore (iii) Google’s preeminence among its users was acquired by virtue of “skill, foresight and industry.”^[1]

With respect to advertisers. This may be a closer call, because the sheer mass of data that Google elicits from endless uses of its search engine, refines its search engine and makes it a more appealing venue for advertisers. In other words, scale is for Google as significant a liability as an asset. Therefore, a court will likely take notice of the potential significant market loss if an advertiser decides to go to a competitor. Nevertheless, Google may assert as a defense that the advertisers retain alternative choices (competing engines) and feasibility to contract with them.

Finally, with respect to Google’s competitors, this may be the toughest dimension to show a good defense. This may also explain why in 2023 the DOJ brought an additional suit against Google for the same conduct but in the advertising technology (adtech) market, which is linked to companies looking to develop products utilizing such technology, including other search engine companies. Here, the exclusionary effect of Google’s contracts may be attempted to be rebutted by pro-user efficiencies (such as an integrated experience, readiness of account access etc). But a court will likely consider the exclusivity agreements under the scope of proportionality, and will likely find that Google’s aggressive behavior was not justified by pro-consumer efficiencies, or a better experience for mobile device users.

Conclusion

U.S. v. Google is a case in which the federal government and the states have allocated substantial resources for more than three years in preparation for the trial. It is unlikely to be decided quickly, and will likely be narrowed down further more, before reaching the core of Google’s exclusionary conduct and its impact. When it does, its procompetitive justifications will be weighted against its exclusionary contracts with Apple, LG, Motorola, and Samsung and others. This will constitute the sharpest point of dispute in the case.

Franchise Agreements: The Case for Limited Non-Compete Clauses

By Alexandros Kazimirov

In early 2023, the Federal Trade Commission proposed a rule under the notice-and-comment process arguing that non-compete clauses constitute an unfair method of competition and therefore violate Section 5 of the Federal Trade Commission Act. The Commission’s intent is to absolve the labor force from binding clauses that impose restrictions on its movement in the market, which in turn harms competition in the country. In the Commission’s enclosed fact sheet, the widespread use of non-competes is highlighted through grossly disproportionate instances, such as the case of a security guard being prevented from getting a job with a new employer by virtue of a two-year non-compete with his previous employer. The Commission’s view is that such over- broad use of non-competes at all levels of employment, cannot be justified to protect trade secrets in light of the fact that some states like California do not enforce such clauses anymore. In the rule proposal, the FTC recognizes that some cases may require deeper inquiry and asks feedback on whether franchisees should be covered by the rule.

Taking as an example a fictional restaurant chain called Big Kahuna Burger (BKB). BKB is in the business of selecting locations, building restaurants, then selling the restaurants and franchising the buyers to allow them to become individually owned and operated BKB restaurants. Let’s assume that all individual BKB restaurants are owned and operated by franchisees. The potential franchisee enters into a standardized franchise agreement which governs many aspects of the franchise operations. BKB franchisees do not receive an exclusive territory, and prospective franchisees are told that franchisees may face competition from other BKB restaurants as well as, of course, from restaurants of other chains. The franchise agreement includes a non-compete clause, which states that: “No employee may seek employment at a different BKB franchise within six months after their termination of employment with their initial BKB franchise.”

It would be an uphill battle to make the case that such a noncompete clause is per se illegal. The reason is because:

(i) it is fairly narrow in scope, i.e. it applies only to BKB franchises and for a limited time,

(ii) does not prescribe price-fixing on its face and

(iii) may retain a pro-competitive effect vis-a-vis other chain restaurants.

The bench is more comfortable in declaring something per se unlawful when the restraint is clearly restrictive on its face. When it is less obvious, the bench may exercise its discretion and review a restraint under the rule of reason theory. It is therefore worth considering whether limited non-competes between franchises can be considered as an unreasonable restraint of trade under a rule of reason analysis.

Would it survive a rule of reason analysis?

In this analysis, the judge would identify two forces: an intrinsic anti-competitive force and an extrinsic pro-competitive force. The intrinsic force concerns the restraint viewed between one franchise and another. In this intrinsic market of employment, the employees are indeed restricted and the clause functions as a brazenly anti-competitive feature because it limits the post-employment options of a franchise employee. The extrinsic force concerns the restraint as viewed between the franchise and the competing restaurants. In this extrinsic market, the employees are not restricted and taken holistically this feature functions more as a pro-competitive feature, because it enhances labor security between franchises and the employees retain an “out” to competing chains.

Next, the analysis would turn to facts and circumstances, i.e. how many opportunities of an “out” the employees of a particular franchise actually have. For example, the judge can select a designated area around a BKB franchise and ask how many competing chains have presence versus BKB franchises. By having an approximate understanding of the market, a judge can determine how broad or narrow the restrictive character of the clause is. For example, if there are more BKB franchises within a designated area than other restaurants, then the intrinsic force of the clause is stronger and therefore it may be interpreted as an unreasonable restraint on competition. Alternatively, if there are more competing chain restaurants than BKB franchises, then the extrinsic force of the clause is stronger and therefore it may be interpreted as a reasonable restraint on competition.

Perhaps weighing the limitations of non-compete covenants and then rewriting them to an acceptable standard is a task that should not burden the courts, some may argue. Traditionally, courts in Delaware and New York (where until recently state law has tolerated non-competes, although state legislatures have indicated to adopt a more hostile stance), have required that restrictions be reasonable in duration, geographic scope and in kind of the business restrained.

However, in Kodiak v. Adams the Court of Chancery admonished parties seeking to have the bench “blue pencil” restrictive covenants to a reasonable and enforceable scope, echoing the growing hesitancy not only to enforce but also to correct non-competes.

Conclusion

Whether the FTC decides to carve out an exception for post-termination non-competes in franchise agreements or moves to include them in its ban, remains unknown so far. Between the hesitation of courts to review non-competes and the lack of flexibility for franchises that a total ban may entail, the former may be the lesser evil.

FTC & DOJ Review of Merger Guidelines 2023

By Alexandros Kazimirov

On July 19, 2023 the Federal Trade Commission (FTC) and Department of Justice (DOJ) released a draft update of the Merger Guidelines. A couple of weeks later, FTC’s Chair Lina Khan and DOJ’s Assistant Attorney General Jonathan Kanter explained the rationale behind some of the changes proposed in a call with the American Economic Liberties Project.

Starting off, both agency heads lauded the feedback shared by workers and employees, people from many sectors, because as enforcers they may have blind spots. They emphasized the importance of a very robust process of public input and encouraged public participation by submitting comments. From their remarks on the Guidelines, there are five distinguishable points worth noting.

Novel theories of economic harm

Kanter mentioned that competition today looks different than in the 1960s. Today markets are more complex and the agencies should be looking at how competition in any particular market functions holistically. Namely, what are the dimensions of competition (i.e. labor, platforms, privacy etc). Once, there is an understanding of how it works in all dimensions, the potential impact is looked into. Besides the implications that this may entail for the scope of merger investigations, with additional burdens of materials produced, it may also suggest a shift in legal reasoning in enforcement cases. That is, not to start by defining the marketplace, but taking a more nuanced approach by accounting all the different interdependencies, and how they may be affected by the merger.

Consequences for antitrust litigation:

In enforcement cases, after the government has defined the market in question and established the concentration, the impact of the merger is examined. The government “establishes a presumption that the transaction will substantially lessen competition. Once such a presumption has been established, the burden of producing evidence to rebut the presumption shifts to the defendants.” FTC v. Staples

The defendants (the firms pursuing the merger) then often submit quantitative and qualitative analysis that the consumer efficiencies resulting from the merger constitute a greater benefit than the risk of lessened competition. Courts have considered pro-consumer efficiencies to be a legitimate defense to the presumption of harm.

“In this expedited appeal, prudence counsels that the court should leave for another day whether efficiencies can be an ultimate defense to Section 7 illegality. We will proceed on the assumption that efficiencies as presented by Anthem could be such a defense under a totality of the circumstances approach.” United States v. Anthem

However, with the government pivoting towards a more nuanced theory where the interests of multiple classes of constituents are taken into account, the focus may not necessarily lie with the consumer-side anymore. This in turn, may negate the rebutting effect that the pro-consumer efficiency defense has.

Pro-labor momentum

Khan followed up saying that rather than focusing on two-dimensional theories of economic harm, they will focus on the impact on all constituencies. For example, the way mergers harm competition not only on the customer side, but also on the supplier side (including for workers). She also described how the guidelines analyze labor markets differently from consumer-oriented markets. Harm to workers can be inflicted by reducing or freezing wages, cutting benefits, or working-schedule unpredictability. The FTC has been consistent in pursuing its pro-labor policies, as for example with the proposed ban on non-competes earlier in 2023.

Lower HHI thresholds but limited resources

The proposed Guidelines include lower thresholds of market share which triggers the presumption of anti-competitive harm. As to whether the market ought to expect an increase in investigations of mergers, Kanter answered that it’s a small fraction of mergers that are investigated. That a lot of the hysteria is overblown, and that the DOJ only challenges what the DOJ considers to be problematic. He said that “best case scenario is that the problematic mergers aren’t coming to us in the first place.” But in the end, he reiterated the need to take into account present market realities. Despite the bravado however, there were 32 investigations in 2021, with 2022 looking more active but less successful. Expecting resource-limited agencies with a losing-streak to substantially pick up the speed may be counterintuitive.

Focus on potential harm to competitors rather than competition

“Taken as a whole, the legislative history illuminates congressional concern with the protection of competition, not competitors, and its desire to restrain mergers only to the extent that such combinations may tend to lessen competition.” Brown Shoe v. United States. The Guideline draft seeks to prevent firms from barring entry of rivals, or tipping the doorman to do so. In other words, reaching a position of strength, whereupon they may exercise their leverage to foreclose entry to competing products, refuse to deal, or take out nascent competitors altogether.

Serial acquisitions

Another point of interest is serial acquisitions, which result in substantial impact in a market when treated in the aggregate. The Guidelines provide that if an individual transaction is part of a firm’s pattern or strategy of multiple acquisitions, the agencies consider the cumulative effect of the pattern or strategy. This pertains to mergers which individually may not be substantial to harm competition, i.e. with a low market share. It also entails a backward-looking review of an existing pattern of transactions.

Conclusion

Throughout the August 10 call, both agency heads reiterated their commitment to operate on the basis of market realities. In his closing remarks, Kanter reaffirmed that it is important to stay rooted in law and in economics. But the agencies have occasionally struggled to convey this to the courts. In 2021, the DOJ challenged the vertical merger between AT&T and Time Warner under Section 7 of the Clayton Act and lost. The courts found the government’s theories of harm to be too detached from economic reality. While the FTC bounced back with the block of Lockheed Martin’s acquisition of Aerojet Rocketdyne in 2021, a series of setbacks followed in 2022. The latest review of Merger Guidelines offers an insight into the motivation of the agencies to pursue their duties administering the Clayton Act rigorously. What remains obscure is whether the federal courts will be persuaded by their novel theories.

Law School Boycotts and the Sherman Act

By Alexandros Kazimirov

In April of 2023, U.S. News & World Report (USNWR) released its first rankings of law schools after a boycott initiated by Yale, Harvard and UC Berkeley Law Schools, which first cut ties with USNWR in November 2022 claiming the magazine used a flawed methodology. Soon after their announcement, the trio’s boycott was joined by several other prominent schools.

Could the boycott of the U.S. News ranking of law schools be a violation of antitrust law?

As a start, it is prudent to address the root causes which led Law Schools to cease dissemination of information related to test scores to the USNWR journal. At the core of this decision are two conflicting motives. Some believe that rankings published by USNWR do not reflect the efforts undertaken by the Law Schools on promoting a diverse body of students. At the same time, Law Schools are not comfortable with the prospect of test score ambiguity (since under the current system, their ranking depends on the test scores they set as requirements for admission) which entails loss of prestige and therefore potentially losing access to other resources as well.

To avoid a compromise between high academic thresholds and bad reputation in diversity efforts, Law Schools may have collectively agreed not to disclose data to the USNWR, which compiles an annual ranking list. Hence, the issue turns on whether this agreement is a restraint that distorts competition under Section 1 of the Sherman Act.

To begin with, the relevant market for this case would be law school education. While Law Schools have agreed to withhold data from USNWR, which is the consumer of this information, the ultimate recipient of the product that USNWR creates is the pool of prospective student applicants. Law Schools are anxious to attract graduates with high test scores, because this perpetuates their reputational preeminence among their peer competitors. Furthermore, Law Schools are highly motivated to project such reputational preeminence because they are also competing for donations and gifts by individual benefactors.

Law Schools may be susceptible to reputational harm but they are not immune from antitrust violations. An agreement not to participate in the ranking list publication resembles a horizontal restraint seeking to curb competition among Law Schools for student admissions. This action is in fact a group boycott, which under U.S. v. General Motors Corp. is unlawful per se, however like in FTC v. Indiana Federation of Dentists, a more deferential judge would emulate the court’s discretion and examine it under the rule of reason instead.

Afterall, like in Indiana Federation of Dentists, the agreement does not prescribe price-fixing on its face, but seeks to withhold from their customer a particular service that they desire. To reach a conclusion on whether this agreement is an unreasonable restraint on competition, courts would avoid embarking on a journey into “the sea of doubt” in accordance with Addyston. Here, this temptation would be to consider whether diversity concerns justify a collective withdrawal from what may be a flawed ranking system.

Instead, a court would focus on the agreement’s pro-competitive or anti-competitive propensities. When considering the pro-competitive and the anti-competitive propensities of this restraint, it leans to the latter. In National Society of Professional Engineers, the court held “the assumption that competition is the best method of allocating resources in a free market recognizes that all elements of a bargain — quality, service, safety, and durability — and not just the immediate cost, are favorably affected by the free opportunity to select among alternative offers.” Similarly, Law Schools have no obligation to submit their data on student admissions to USNWR, but they cannot collectively agree to foreclose the possibility of doing so, in an attempt to prevent competition among themselves, and reduce the risk of either reputational harm or lower revenues from donations.

Like in Indiana Federation of Dentists, “a refusal to compete with respect to the package of services offered to customers impairs the ability of the market to advance social welfare by ensuring the provision of desired goods and services to consumers at a price approximating the marginal cost of providing them.” In this instance, the consumer – USNWR – would be less informed about the test scores of the admitted students. In turn, this would result in greater ambiguity in the decision-making process of admissions, i.e. what criteria are used and to what effect. In other words, less information leads to less competition.

Without a feature which enhances its pro-competitive efficiencies, this restraint cannot be viewed as legal. It is plausible that Law School may dislike the rankings publication on the premise that it distorts or misinforms their efforts on attracting a diverse body of students. That does not mean that law schools are precluded from withdrawing from the rankings publication on their own initiative; rather it confirms that to do it in a concerted manner would distort competition.

In sum, concerns of diversity and inclusion may be a legitimate worry, but not one that makes a restraint on competition reasonable. Like the court’s reasoning in National Society of Professional Engineers, “we may assume that competition is not entirely conducive to ethical behavior, but that is not a reason, cognizable under the Sherman Act, for doing away with competition.”

What if there was never an agreement to begin with?

Rather, a case where Yale took the lead, and the rest followed in what constitutes parallel conduct. Parallel conduct can be explained as conduct which is sound on its own, regardless of actions of competitors, or it can be so suspicious that it raises an inference of a conspiracy.

An express agreement of a restraint of a trade is the best evidence one can hope for, however direct evidence of conspiracy is not a sine qua non. Circumstantial evidence can establish an antitrust conspiracy under Interstate Circuit. Circumstantial evidence can be shown through price exchange, regular peer contacts, stating shared concerns so that “knowing that concerted action was contemplated and invited, the distributors gave their adherence to the scheme and participated in it.”

The first requirement is an atypical behavior. By initiating the boycott, Yale is undertaking a highly risky move, which if not followed by others could be harmful for its reputation. Is it a rational decision? Would Yale commit to its decision if not sure that others would follow? Well, others joined the boycott. Perhaps, the bench could interpret this as a series of reasonable independent initiatives or someone more inquisitive could infer a conspiracy. But inference alone is not dispositive. It simply gives the bench reason to consider circumstantial evidence. Therefore, it would ask whether the law school deans consulted among themselves, exchanged information on resources allocated to diversity or formulated appropriate courses of action. Such evidence would be holistically considered before concluding whether there was a tacit accord to partake in a boycott. And if the bench concluded in the affirmative, it would find it illegal per se or apply the aforementioned rule of reason analysis, where absent any pro-competitive features, a conspiracy to distort competition would be a violation of the antitrust law.

Intellectual Property

United States

AI, Face Swapping, and Right of Publicity

By Marie-Andrée Weiss

Last April, several plaintiffs filed a putative class action against NeoCortext, Inc., the developer of the Reface face swapping application, alleging that the application infringed their right of publicity.

NeoCortext moved to dismiss the complaint, claiming that plaintiffs’ right of publicity was preempted by the Copyright Act and barred by the First Amendment. NeoCortext also moved to strike the complaint, claiming that the suit was a strategic lawsuit against public participation (SLAPP) aiming at “gagging a novel application that enables users to engage in creative activities that are protected by the First Amendment.”

On September 5, 2023, U.S. District Judge Wesley L. Hsu denied both motions.

The case is Kyland Young v. NeoCortex Case 2:23-cv-02496-WLH-PVC.

The Reface app

Neocortext developed Reface, a smartphone application using an artificial intelligence algorithm which allowed users to replace their faces in photographs and videos with the faces of celebrities (“face swap”), to place their faces into scenes and movies and to “mix [their] face[s] with a celebrity.”

Users were able to search for their favorite characters or individuals in the catalog of images, movie and show clips, which was compiled from several websites, such as mybestgif.com, https://tenor.com/, Google Video, and Bing Video. Among the individuals featured in the catalog was one of the plaintiffs, Kylan Young, finalist of the 23rd Big Brother show on CBS.

Users  could then upload a photograph featuring one or more human beings, and the app “swapped” the faces with the faces of individuals featured in the images or clip chosen by the user from Reface’s catalogue. NeoCortext offered a free version of the services, where the “face swap” image or video was watermarked with the Reface logo. The complaint referred to these watermarked images and clips as “Teaser Face Swaps.” A paying subscription to the app allowed the user to remove the watermark.

Does the app infringe plaintiff’s right of publicity?

The complaint alleged that the app allowed users to recreate Mr. Young’s scenes from Big Brother, but that NeoCortext never asked for his consent nor paid him any royalties and thus profited from Mr. Young’s likeness and that defendant used the likeness of plaintiffs in violation of California’s right of publicity “to pitch its product for profit.” Plaintiff argued that  Teaser Face Swaps were “essentially ads intended to entice users to buy PRO subscriptions, and the paid PRO version of the applications makes money by including Californians in its library of content.”

California Right of Publicity Law

California recognizes a right of publicity at common law and also by statute, California Civil Code § 3344, which prevents the use without prior consent of a person’s name, voice, signature, photograph or likeness, in products, merchandise or goods, to advertise, sell, or solicit the purchase of goods or services. 

To succeed, a plaintiff must allege  that (1) the defendant’s used the plaintiff’s identity; (2) appropriated the plaintiff’s name or likeness to defendant’s advantage, commercially or otherwise; (3) defendant did not consent;  and (4) injury resulted from this unauthorized use (see for instance Fleet v. CBS, Inc. at  1918).

The two Anti-SLAPP steps.

In its motion to strike the case, NeoCortext argued that the app allowed its users to create “humorous and sometimes absurd new works for personal use” and that “[t]his is exactly the type of creative activity that the First Amendment protects and that the right of publicity does not.”

There are two steps in an anti-SLAPP analysis, the second step being equivalent of the standard used by courts to evaluate a motion to dismiss.

First step:

The first step under California Anti- SLAPP law, Cal. Civ. Proc. Code § 425.16, was for NeoCortext to show that its use of Mr. Young’s image was made “in furtherance of [NeoCortext’s] right of petition or free speech… in connection with a public issue. Such speech can be conduct, including “all conduct in furtherance of the exercise of the right of free speech” (Lieberman v. KCOP Television, Inc., at  166).

Judge Hsu reasoned that the conduct at the basis of Mr. Young’s complaint was the inclusion of his image in the app, allowing users to create a new image.  As such, it was the users who exercised their freedom of speech, not NeoCortext. Because the app is a tool that users can use to exercise their free speech rights, NeoCortext’s use of plaintiff’s image in the app was conduct taken in furtherance of users’ exercise of free speech.

Such speech is connected with a public issue under the test used by California courts as it is: a (1) statement concerning a person or entity in the public eye (Mr. Young); (2)  a conduct that could directly affect a large number of people beyond the direct participants; (3) or a topic of widespread public interest (“the use of technology to alter images and videos of individuals in a way that makes them look realistic” is such topic).

NeoCortext had shown that its conduct is in furtherance of the right of free speech made in connection with a public issue, thus satisfying its burden on the first step of the anti-SLAPP analysis.

Second step:

Plaintiff therefore then carried the burden to show “a probability of prevailing on the claim”, the second step required by California Anti-SLAPP law, identical to the standard for the motion to dismiss, and it did so, leading Judge Tsu to deny both motions.

NeoCortext had argued, unsuccessfully as we will now see, that the Copyright Act and the First Amendment preempted the right of publicity claim.

Copyright Act does not preempt the right of publicity claim

NeoCortext had argued that, if a right of publicity claim is entirely based on the display, reproduction or modification of a work protected by copyright, the claim is preempted by the Copyright Act.

Section 301 of the Copyright Act preempts state laws equivalent to the exclusive copyright rights as detailed by Section 106 of the Copyright Act.

The Ninth Circuit uses a two-part test to determine if a state law claim is preempted by the Copyright Act :

• (1) the subject matter of the claim must fall within the subject matter of copyright as described in Section 102 and Section 103 of the Copyright Act and;
• (2) the rights asserted under state law must be equivalent to the exclusive right on copyrights works (such as the right to reproduce or the right to prepare derivative works) rights contained in Section 106 of the Copyright Act (see Maloney v. T3Media, Inc., at 1010).

NeoCortex had claimed that Plaintiff’s claim was within the subject matter of copyright, as the images and clips in Neocortext’s catalog were protected by copyright.

In Maloney, the Ninth Circuit Court of Appeals held :

“that a publicity-right claim is not preempted when it targets non-consensual use of one’s name or likeness on merchandise or in advertising. But when a likeness has been captured in a copyrighted artistic visual work and the work itself is being distributed for personal use, a publicity-right claim interferes with the exclusive rights of the copyright holder, and is preempted by section 301 of the Copyright Act.” (Maloney, at 1011, our emphasis).

NeoCortex’s argument relied further on Maloney which held that :

“…where a likeness has been captured in a copyrighted artistic visual work and the work itself is being distributed for personal use, a publicity-right claim is little more than a thinly disguised copyright claim because it seeks to hold a copyright holder liable for exercising his exclusive rights under the Copyright Act.” (Maloney, at 1016).

First part of the Ninth Circuit test: Plaintiffs’  right of publicity claim do not fall within the subject matter of copyright

Nothing that the Copyright Act protects ownership of photographs, but that it does not protect the exploitation of a person’s  likeness, “even if it is embodied in a photograph”, citing the Ninth Circuit decision in Downing v. Abercrombie & Fitch, Judge Hsu found that “[plaintiff]’s right of publicity claim does not fall within the subject matter of copyright”. Judge Hsu  distinguished the case from Maloney, where a photograph of the plaintiff, protected by copyright,  had been sold. In contrast, the use of Mr. Young’s likeness was outside of the original work protected by copyright as it was used to create a product containing the plaintiff’s image.  As plaintiff’s claim did not fall under the subject matter of copyright, it was not preempted by the Copyright Act.

Second part of the Ninth Circuit test: State law rights asserted are not equivalent to Section 106  rights

Judge Hsu also found that the second factor of the test failed, because Section 106 of the Copyright Act does not give  the owners of the photographs the right to use plaintiff’s name and likenesses to advertise the free version of the app and to induce users to buy the subscription. Plaintiff was “not seeking to “merely” restrict the reproduction or distribution of the original photographs/works, as the plaintiffs in Maloney ….”

The rights asserted by plaintiff were not equivalent to the rights conferred by the Copyright Act to the owners of the photographs from the app catalog. Under the two-part test used by the Ninth Circuit, the claim was not preempted by the Copyright Act.

The First Amendment does not preempt the right of publicity claim

NeoCortext had also argued that the First Amendment preempted the claim, as users used the app to create “their own unique, sometimes humorous and absurd expressions” which are protected by the First Amendment. NeoCortext further argued that the photos and clips thus created had “creative and aesthetic value” and that they were “new works … distinct from the originals”.

California courts apply the “transformative use” test to balance right of publicity and First Amendment, detailed by the California Supreme Court in Comedy III Productions v. Gary Saderup, Inc. (at 142):

“In sum, when an artist is faced with a right of publicity challenge to his or her work, he or she may raise as affirmative defense that the work is protected by the First Amendment inasmuch as it contains significant transformative elements or that the value of the work does not derive primarily from the celebrity’s fame.” (Our emphasis).

NeoCortext had to show that its use was transformative as a matter of law. Judge Hsu found it had not done so, noting that plaintiff’s face “is the only thing that change in the end product” and that the body is sometimes unchanged, citing  Hilton v. Hallmark Cards, where the Ninth Circuit  found that a greeting card featuring the likeness of Paris Hilton, arguably more transformative than the swap images created the app, was not transformative enough to entitle the defendant to a First Amendment affirmative defense as a matter of law.

What is next?

On September 8, NeoCortex filed an appeal to the U.S. Court of Appeals for the Ninth Circuit.

There have already been several complaints alleging that an AI-powered product or service is infringing the copyright of authors whose works have been used to train the data models, but Young v. NeoCortext is one of the first cases were a product or service triggered by AI is allegedly infringing a right to publicity.

As such it is worthy of following further. To be continued…

Other Developments

United States

SEC’s First Enforcement Case under Regulation Best Interest

By Alexandros Kazimirov

On March 13, 2023, Judge Wright issued an order granting in part and denying in part a motion to strike brought by the Securities & Exchange Commission (SEC) in the first case pertaining to the enforcement of Regulation Best Interest (Reg BI). Reg BI marks the departure from the former suitability rule to the new enhanced best interest rule. To this end, it is prudent to delineate the scope of enforcement the SEC is seeking under the new rule.

Summary of facts

The SEC v. Western International Securities, Inc., (Western) is a case in the Central District of California, first filed on June 15, 2022. The case scrutinizes the conduct of Pasadena-based advisory firm Western International Securities, Inc. and five of its registered representatives under the new best interest rule.

According to the complaint, Western’s brokers recommended the sale of debt securities called L Bonds, which had no market-out and were unrated to several retail customers. The issuer of these bonds had described the bonds as high-risk and speculative securities in the prospectus, but Western’s brokers recommended them to investors with a conservative risk tolerance, investment objectives that did not include speculation, limited investment experience and limited liquid net worth.

The SEC alleges that this disproportionate matching violates the obligation of care that the brokers owe to their clients. The brokers did not exercise the reasonable care and diligence to assess and tailor the securities to their customers’ risk-sensitive profiles, and had no reasonable basis to believe their recommendation was in their clients’ best interests.

Furthermore, the complaint also refers to Western’s inadequate compliance mechanism vis-a-vis the incorporation of the best interest rule into the firm’s practice. Namely, Western did not undertake any substantial effort of explaining to its broker-dealers how to be compliant with the principles under the new regulation.

The SEC is seeking permanent injunctions, disgorgement and prejudgment interest, and civil penalties against both Western and the broker-dealers involved.

Best Interest Rule

The literature on Reg BI mentions that the “SEC has no definition of ‘best interest’. Instead, whether a broker-dealer has acted in a retail customer’s best interest is based on an objective assessment of the facts and circumstances of how the broker- dealer has satisfied the four component obligations of Regulation Best Interest at the time the recommendation is made.”^[2]

Generally, for a dealer-broker who sells products to retail investors, the duty of good faith and fair dealing implies several things. First a broker must ascertain the nature of the product, which includes both its features and its purpose, such as the targeted clientele. Then the broker must lay the foundation, i.e. the reasonable basis, on which the product may be recommended to a client who is suitable for the purchase.

If the broker’s relationship predates the recommendation, the broker must have disclosed all material facts as to the relationship terms to the client. Last, the broker’s conduct must be consistent; it ought to be happening in a coherent and organized manner which adheres to the principles of the best interest rule. Therefore, when a broker acts as a diligent, loyal seller in good faith and the terms of the sale are reasonable, the broker has acted in the client’s best interest.

Specifically, Reg BI lays out four component obligations:

Disclosure obligation: a broker-dealer, before or at the time of the recom- mendation, must provide the retail customer, in writing, full and fair disclosure of all material facts as to the scope and terms of its relationship with the retail customer and all material facts relating to conflicts of interest that are associated with a recommendation. The SEC did not identify the duty of disclosure to be an issue in Western.

Care obligation: Reg BI’s care obligation requires a broker, dealer, or associated person of a broker or dealer, in making a recommendation of a securities transaction, to exercise reasonable diligence, care, and skill to (a) understand the potential risks, rewards, and costs associated with the recommendation and to (b) have a reasonable basis to believe the recommendation is in the best interests of that customer, based on the customer’s investment profile and the potential risks, rewards, and costs associated with the recommendation.

This component includes a two-part test: first comprehension of the product that is sold (diligence) and second matching the product to the needs and profile of the client (proportionality). In Western, the SEC asserts that the brokers were unaware of the issuer’s transactions, whether there was collateral security for the L Bonds and lacked appreciation of the risk that a purchase of such securities entailed. For the SEC issues pertinent to the product’s collateral can hardly be seen as of secondary nature. In other words, Western’s brokers failed the first prong.

The complaint then addresses what the SEC’s claims to be a mismatch. Namely that the L Bonds were securities bearing substantial risk and were intended for clients with substantial financial resources and no need for liquidity. But Western’s clients did not meet such criteria. The customers were mostly risk averse retirees who allocated between a tenth and a third of their cash savings in L Bonds. Therefore, according to the SEC, Western’s brokers failed the second prong, too.

Conflict of interest obligation: a broker-dealer must establish, maintain and enforce policies and procedures reasonably designed to identify, monitor and mitigate (or eliminate, if possible) conflicts of interest. Conflict of interest is not an issue in Western.

Compliance obligation: a broker-dealer must establish, maintain and enforce policies and procedures reasonably designed to achieve compliance with Reg BI.

Compliance under the best interest rule provides that broker firms should undertake the effort of explaining the effect of the principles of the regulation to their brokers in practice, and pursue such an effort consistently. A firm should not just reiterate its commitment to Reg BI or merely restate SEC’s guidance which fails to reach its broker-dealers, as in Western.

The complaint alleges that Western had a dysfunctional best interest compliance policy. Western’s Chief Compliance Officer (CCO) prepared due diligence reports which were out of reach for the dealer-brokers, and the latter were untrained in the latest offering of L Bonds, class of 2020-2021. This reflects a bifurcated system of due diligence, which according to the SEC, appears to have been largely asynchronous between the firm and the representative brokers.

With the brokers having no access to due diligence reports, their under- standing of the potential risk margin was grossly inadequate for the purpose of recommending the securities to their clients. And with the firm not encou- raging its brokers to stay up-to-date with the issuer’s product, it facilitated the lack of appreciation of its brokers and the eventual mismatch with the retail investors.

FINRA’s Suitability Rule

The question then is whether the recommendation of the debt securities would be adequate under the previous suitability standard. The test requires a broker to reasonably believe that a recommendation would be a good fit for the client, based on previous due diligence of the client’s profile. The suitability test lays out three obligations:

Reasonable-basis suitability: the reasonable-basis obligation required broker-dealers (through reasonable diligence) to understand the risks and rewards involved with a particular investment and to have a reasonable basis to believe that an investment was suitable for at least some investors.

Customer-specific suitability: the customer-specific obligation required broker-dealers to have a reasonable basis to believe an investment was suitable for a specific customer based on that customer’s investment profile.

Quantitative suitability: the quantitative suitability obligation required broker- dealers to reasonably believe that a series of individually suitable recom- mendations, when considered in con- cert, would not become excessive and therefore unsuitable for an investor.

In Western, the SEC argues that the retail investors were not fairly sophisticated investors, who had moderate risk tole- rance. Some indicated that they wanted to “earn income safely without risking [the] principal.” The investors were pensioners who had allocated a substantial portion of their liquid net-worth to purchase the issuer’s L Bonds, and relied on their brokers’ judgment.

In forming such judgment, the brokers did not take into account (i) important transactions pertaining to the issuer, (ii) warnings about the high-risk nature of the product in the issuer’s prospectus, (iii) that L Bonds were not directly collateralized by life insurance policies held by the issuer, and (iv) training as to the particular features of the latest issue of L Bonds.

From the aforementioned facts, a reasonable broker could not have proceeded to match the L Bonds, as suitable products for Western’s clients. Where the match between the retail customer profile and the recommendation appears less reasonable, like in Western, it is more important for the broker to establish their reliance on a reasonable belief that the recommendation was in the best interest of the retail customer.

However, such reasonable belief cannot be inferred under any of the suitability obligations, because the SEC alleges that the brokers lacked rudimentary understanding of the product’s features to form the reasonable basis by which they could recommend it to their client. Furthermore, the description of the products in the issuer’s prospectus was opposite to the investment profile that the clients maintained with Western. If the clients were not the typical target of the debt securities, and the brokers knew that, they could not have reasonably found them to be suitable for their clients.

Finally, the unsuitable nature of the products would not be reversed if the sale was treated in the aggregate or otherwise. Therefore, the facts and circumstances of Western, as applied under the suitability obligations of reasonable-basis, customer- specific, and quantitative suitability do not arrive to a different conclusion than under the best interest rule.

Conclusion

In SEC v. Western International Securities, Inc., the standard of conduct under scrutiny is fairly consistent with the previous threshold of the suitability rule. Namely that the recommendation of a security to retail investors be a reasonably proportionate fit to the investor’s profile. In other words, the SEC has not materially expanded the reach of the best interest rule, so far.

An enforcement agency would still find the recommendations described in Western to be below the threshold set by the suitability requirements, previously held. However, despite the fact that there is no radical departure in scope from the former rule, the sharpest point of contention rests less on the conduct of the broker-dealers and more on the perceived shortcomings of Reg BI.

In Western, the defendants claimed the affirmative defense of lack of fair notice, which pertains to the definitional clarity of Reg BI and asks the court to consider “whether the challenged statute is unconstitutionally vague as applied to the particular facts at issue such that the challenging party did not have sufficient notice that his or her conduct would be a violation of the statute.” Judge Wright refrained from reaching a conclusion on the legal determination of fair notice before the factual dispute is resolved, citing the Ripple case. It is evident however, that this will be the main issue.

European Union

Cyberstalking and Online Platforms’ Due Diligence in the EU Digital Services Act

By Irene Kamara

Cyberstalking: a pattern of abusive online behaviours

Cyberstalking, the act of using electronic communication devices to create a criminal level of intimidation, harassment, and fear in one or more victims,^[3] is a form of – usually gender-based- cyberviolence, with immense impacts on the physical and mental well-being of the victim. The Council of Europe Istanbul Convention on violence against women and children defines stalking as “the intentional conduct of repeatedly engaging in threatening conduct directed at another person, causing her or him to fear for her or his safety.”^[4] The characteristic of cyberstalking is the repeated nature of the online harassment. It constitutes a pattern of behaviour, rather than one isolated incident.^[5] Because of this aspect, while the victim may feel a continuous threat, classifying different events by a single or multiple offenders as one cyberstalking offence and prosecuting it, runs into several evidentiary obstacles. Such an evidentiary obstacle is that the victim needs to maintain records of the different events over the course of an extended period that amounts to the cyberstalking offence. Where punishable, cyberstalking usually falls under criminal law provisions of harassment, especially in jurisdictions that have signed and ratified the Istanbul Convention of the Council of Europe. However, regulatory approaches targeting the offender are not the only strategy to mitigate cyberstalking as a phenomenon. Online platforms such as social media platforms offer de facto a means that facilitates cyberstalking, since offenders use social media platforms to engage in unwanted communication such as threats against one or more victims or publicise defamatory or image-based abusive material. Several of the most popular platforms have adopted their own community standards on accepted behaviour. For example, Meta has a policy in place on bullying and harassment,^[6] where inter alia the platform commits to “remove content that’s meant to degrade or shame, including, for example, claims about someone’s sexual activity.” Those policies however are largely voluntary measures, and their appropriateness is often not reviewed by external state actors, such as an independent supervisory authority.

Cyberstalking and the EU Digital Services Act

Since 2022, the EU has a new Regulation in place assigning a range of responsibilities to online platforms, such as Meta, to identify and take down illegal content including cyberstalking. The Digital Services Act (‘DSA’)^[7] aims at providing harmonised EU rules for a “safe, predictable and trusted online environment”,^[8] by inter alia establishing rules on due diligence obligations for providers of intermediary services. The DSA modernised some of the provisions of the 2000 e-Commerce Directive^[9] and reinstated some others, such as the provision clarifying there is no general obligation for providers of intermediary services to monitor the information in their services, nor to engage into active fact-finding to establish whether an illegal activity takes place abusing their services.^[10]

Despite the absence of a general monitoring obligation, providers of intermediary services are subject to several obligations in order to ensure the online safety and trust of the users of their services.^[11] Those due diligence obligations, explained in the next section, are centered around the concept of illegal content. The DSA, defines in its Article 3(h) illegal content as “any information that, in itself or in relation to an activity, including the sale of products or the provision of services, is not in compliance with Union law or the law of any Member State which is in compliance with Union law, irrespective of the precise subject matter or nature of that law.” The concept of content is thus very broad meaning any information, ‘products, services and activities’^[12] and whether this content is illegal is determined by examining other EU or Member State law. Once thus information shared, publicized, transmitted, stored that is infringing EU or national Member State law, the due diligence framework established in the DSA is applicable to the service provider of intermediary services. Recital 12 DSA provides additional interpretational clarity as per the parameters and examples of illegal content, since applicable rules might render the content itself illegal or this might be rendered illegal because it relates to illegal activities. Examples include the sharing of image-based sexual abuse of children material (CSAM), hate speech or terrorist content, and online stalking (cyberstalking). As a result of this broad definition, even acts or information that are not as such illegal, but relate to the illegal activity of cyberstalking, would also qualify as illegal content, and would be subject to the DSA. This is an important step towards regulating cyberstalking, and essentially limiting the single acts of the cyberstalker causing nuisance or harassment to the victim(s) and other related targets of the offence, such as the friendly, family, or work environment of the victim(s).

The DSA due diligence framework: placing the responsibility on online platforms?

The e-Commerce Directive already provided an obligation for information society service providers to remove or disable access to information, when obtaining knowledge of an illegal activity.^[13] The DSA develops a due diligence framework, which involves service providers undertaking action in a reactive manner (e.g. once a report is filed towards an online platform about an abusive image), but also in a proactive manner. The due diligence framework ensures that the service providers, and especially large online platforms, have assessed the systemic risks from the design and the functioning of their services.^[14] The due diligence framework comprises of rules relating to transparency, cooperation with law enforcement and judicial authorities, and proactive measures against misuse of the offered services. In terms of proactive measures, very large online platforms must put in place mitigation measures tailored to systemic risks and adapt their moderation processes,  in particular in cases of cyberviolence, which includes cyberstalking. The risk of dissemination of CSAM is – according to Recital 80 DSA – one of the categories of such systemic risks. The mitigation measures include the expeditious removal or disabling access to the illegal content, and adapting the speed and quality of processing notices (Art. 35(1)(c) DSA). In terms of transparency, specifically for online platforms, the DSA imposes strict reporting rules as regards the use of automated moderation tools, including specification of error rates and applied safeguards,^[15] but also detailed reporting of the number of suspensions of provision of services due to misuse.^[16] As regards cooperation with law enforcement and judicial authorities, all hosting providers must notify the competent authorities of a suspicion that a criminal offence threatening an individual’s safety or life is taking place. The notification threshold is quite low, since Art. 18(1) DSA requires not proven illegal behaviour, but a suspicion that such a behaviour takes place. This means that in cases of cyberstalking, any act pointing the service provider at the direction of a potential of repeated threats directed towards an individual directly or indirectly via friends, family, or colleagues would require a report to the law enforcement authority.

Next steps

The DSA entered into force in 2022, but starts applying early 2024, since the EU legislator provided a grace period to service providers subject to the scope of the DSA to adapt to the new set of obligations. While it should be expected that hate speech, CSAM, and copyright infringing material, will -at the first period of the DSA application monopolise the focus of platforms and the related complaints and reports- the DSA will also be tested as a regulatory instrument against cyberstalking and the role of intermediaries, e.g. in this case online platforms, in combatting such an abusive online behaviour.

EU Adoption of DAC 8 – Mandatory Exchange of Information between Tax Authorities on Crypto Assets

By Amedeo Rizzo

On the 17th of October 2023, the Council of the European Union approved Directive DAC 8 on administrative cooperation (Press Release), introducing significant modifications related to the communication and automatic exchange of information regarding proceeds from operations in crypto-assets and information on advance tax rulings for high-net-worth individuals. With this directive, the EU, considering the new opportunities brought about by digitalization, aims to expand the scope of the obligation for automatic exchange of information, fostering a higher degree of administrative cooperation among tax administrations.

Crypto assets definition and tax problems

The term crypto asset refers to a digital representation of value that relies on a cryptographically secured distributed ledger to validate and secure transactions^[17]. This mechanism establishes a tamper-resistant record of transactions within the asset without the need for a central authority. The challenge in categorizing assets within this broad class arises from ongoing innovation and the diverse range of services that specific assets can offer. Distinguishing these assets for tax purposes is complex due to these factors.

However, a fundamental tax-relevant dimension that aids in their characterization is the distinction between their use for investment purposes and as a means of payment. At one end of the spectrum are “security token,” which essentially serve as digital representations of traditional financial or other assets. An example includes “Non-fungible tokens” (NFTs), which are cryptographically protected representations of unique assets, such as works of art. Conversely, central bank digital currencies (CBDCs), might be considered to be more similar to fiat currency in digital form. While some national governments remain cautious about their adoption, the prevailing expectation is that the issuance of CBDCs will become widespread over time^[18].

The primary impediment in the taxation of crypto assets stems from their inherent “anonymous” nature, wherein transactions employ public addresses that prove exceptionally challenging to associate with individuals or entities. This characteristic introduces a heightened susceptibility to tax evasion, placing the onus on tax authorities to address implementation challenges effectively.

When transactions occur through centralized exchanges, the challenge becomes more manageable as these exchanges can be subjected to standard know your customer (KYC) tracking rules and potential withholding taxes.

Background and content

On December 7, 2021, the Council, in its report to the European Council regarding tax matters, communicated its anticipation that the European Commission would present a legislative proposal in 2022 for the additional amendment of Directive 2011/16/EU on administrative cooperation in taxation (DAC).

This proposed amendment specifically pertained to the exchange of information regarding crypto-assets and tax rulings applicable to individuals with substantial wealth. According to the Council, it was imperative to fortify the stipulations of Directive 2011/16/EU pertaining to the information to be reported or exchanged to accommodate the evolving landscape of diverse markets and, consequently, to effectively address identified instances of tax fraud, tax evasion, and tax avoidance, by facilitating effective reporting and exchange of information.

In light of this objective, the Directive encompasses, among other aspects, the most recent revisions to the Common Reporting Standard (CRS) of the OECD. Notably, this includes the incorporation of provisions pertaining to electronic money and central bank digital currencies (CBDCs) delineated in Part II of the Crypto-Asset Reporting Framework and Amendments to the Common Reporting Standard, endorsed by the OECD on August 26, 2022.

Moreover, the Directive extends the purview of the automatic exchange of information concerning advance cross-border rulings to encompass specific rulings concerning individuals. In particular, it includes in the scope of the current regulation the rulings involving high-net-worth individuals, as well as provisions on automatic exchange of information on non-custodial dividends and similar revenues.

Additionally, the Directive enhances the regulations governing the reporting and communication of Tax Identification Numbers (TIN). The objective is to streamline the identification process for tax authorities, enabling them to accurately identify pertinent taxpayers and assess associated taxes. Additionally, the Directive seeks to modify provisions within the DAC concerning penalties imposed by Member States on individuals who fail to comply with national legislation related to reporting requirements established in accordance with the DAC.

This approach is adopted to ensure uniformity and coherence in the application of these provisions across Member States.

Problems addressed by the Directive

The bottom line of the DAC 8 revolves around the imperative of instituting mandatory reporting for crypto-asset service providers falling within the ambit of the Markets in Crypto-Assets (MiCA) Directive. Additionally, all other crypto-asset operators offering services to residents of the EU are required to comply. Non-EU operators must undergo registration in a Member State to adhere to DAC 8 regulations, ensuring the reporting of pertinent information. This strategic approach equips tax authorities of Member States with the requisite tools to monitor income generated from crypto assets by EU users and implement necessary measures to ensure tax compliance.

The reporting mechanism entails three sequential steps. Initially, crypto-asset service providers collect information of the transactions subject to reporting by their users. Subsequently, the providers submit the compiled information to the competent tax authority of their Member State (for EU providers) or the competent authority of the Member State of registration (for non-EU providers). Lastly, the competent tax authority transmits the reported information, inclusive of the TIN of the reported users, to the competent authorities of the users’ respective Member States of residence.

The Directive also emphasizes reporting requirements concerning reportable users and crypto assets. Reportable users are mandated to furnish their:

• complete name;

• address;
• Member State of residence;
• date and place of birth;
• TIN.

Reportable crypto assets are to be identified by their complete name and the aggregate gross amount paid or the aggregate fair market value.

Reporting crypto-asset service providers are obligated to obtain a self-certification from users, encompassing information crucial for determining the user’s tax residence, such as full name, date of birth, residence address, and TIN. The proposal allows a substantial degree of discretion in evaluating the reliability of this self-certification, permitting providers to verify information using alternative sources, including their own customer due diligence procedures, in case of doubts. If a user accesses the platform through a Member State’s digital identity system, the provider is exempt from collecting certain information but is still required to obtain the user’s full name, the identification service used, and the Member State of issuance.

The Directive incorporates provisions facilitating the effective implementation of the proposed measures, including mechanisms for enforcing compliance by non-EU crypto-asset operators with EU resident users. In instances where non-EU operators fail to comply with reporting obligations due to a lack of registration in a Member State, the DAC 8 grants Member States the authority to employ effective, proportionate, and dissuasive measures to ensure compliance, potentially encompassing measures that may prohibit the operator from operating within the EU as a last resort (Article 8ad).

Conclusion

In summary, the recently approved DAC8 emerges as one of the needed responses to the evolving landscape of crypto assets, acknowledging some of the inherent challenges in taxation posed by their anonymous nature and the dynamic innovation within this domain.

By bridging the information gap and enhancing reporting mechanisms, DAC 8 empowers tax administrations to monitor and enforce compliance, thus mitigating some of the potential tax risks associated with crypto assets and tax rulings. The Directive, with its comprehensive approach and emphasis on international cooperation, is a critical step towards achieving transparency in the taxation of these emerging financial instruments.

Large Language Models and the EU AI Act: the Risks from Stochastic Parrots and Hallucination

By Zihao Li^[19]

With the launch of ChatGPT, Large Language Models (LLMs) are shaking up our whole society, rapidly altering the way we think, create and live. For instance, the GPT integration in Bing has altered our approach to online searching. While nascent LLMs have many advantages, new legal and ethical risks^[20] are also emerging, stemming in particular from stochastic parrots and hallucination. The EU is the first and foremost jurisdiction that has focused on the regulation of AI models.^[21] However, the risks posed by the new LLMs are likely to be underestimated by the emerging EU regulatory paradigm. Therefore, this correspondence warns that the European AI regulatory paradigm must evolve further to mitigate such risks.

Stochastic parrots and hallucination: unverified information generation

One potentially fatal flaw of the LLMs, exemplified by ChatGPT, is that the generation of information is unverified. For example, ChatGPT often generates pertinent, but non-existent academic reading lists. Data scientists claim that this problem is caused by “hallucination”^[22] and “stochastic parrots”.^[23] Hallucination occurs when LLMs generate text based on their internal logic or patterns, rather than the true context, leading to confidently but unjustified and unverified deceptive responses. Stochastic parrots is the repetition of training data or its patterns, rather than actual understanding or reasoning.

The text production method of LLMs is to reuse, reshape, and recombine the training data in new ways to answer new questions while ignoring the problem of authenticity and trustworthiness of the answers. In short, LLMs only predict the probability of a particular word coming next in a sequence, rather than actually comprehending its meaning. Although the majority of answers are high-quality and true, the content of the answers is fictional. Even though most training data is reliable and trustworthy, the essential issue is that the recombination of trustworthy data into new answers in a new context may lead to untrustworthiness, as the trustworthiness of information is conditional and often context-bound. If this precondition of trustworthy data disappears, trust in answers will be misplaced. Therefore, while the LLMs’ answers may seem highly relevant to the prompts, they are made-up.

However, merely improving the accuracy of the models through new data and algorithms is insufficient, because the more accurate the model is, the more users will rely on it, and thus be tempted not to verify the answers, leading to greater risk when stochastic parrots and hallucinations appear. This situation, where an increase in accuracy leads to higher reliance and potential risks, can be described as the ‘accuracy paradox’. The risk is beyond measure if users encounter these problems in especially sensitive areas such as healthcare or the legal field. Even if utilizing real-time internet sources, the trustworthiness of LLMs may remain compromised, as exemplified by factual errors in new Bing’s launch demo.

These risks can lead to ethical concerns, including misinformation and disinformation, which may adversely affect individuals through misunderstandings, erroneous decisions, loss of trust, and even physical harm (e.g., in healthcare). Misinformation and disinformation can reinforce bias,^[24] as LLMs may perpetuate stereotypes present in their training data.^[25]

The EU AI regulatory paradigm: Advanced Legal intervention required

The EU has already commenced putting effort into AI governance. The AI Act (AIA) is the first and globally most ambitious attempt to regulate AI. However, the proposed AIA, employing a risk-based taxonomy for AI regulation, encounters difficulties when applied to general-purpose LLMs. On the one hand, categorizing LLMs as high-risk AI due to its generality may impede EU AI development. On the other hand, if general-purpose LLMs are regarded as chatbots, falling within a limited-risk group, merely imposing transparency obligations (i.e., providers need to disclose that the answer is generated by AI) would be insufficient.^[26] Because the danger of parroting and hallucination risks is not only related to whether users are clearly informed that they are interacting with AI, but also to the reliability and trustworthiness of LLMs’ answers, i.e., how users can distinguish between truth and made-up answers. When a superficially eloquent and knowledgeable chatbot generates unverified content with apparent confidence, users may trust the fictitious content without undertaking verification. Therefore, the AIA’s transparency obligation is not sufficient.

Additionally, the AIA does not fully address the role, rights, or responsibilities of the end-users. As a result, users have no chance to contest or complain about LLMs, especially when stochastic parrots and hallucination occur and affect their rights. Moreover, the AIA does not impose any obligations on users. However, as aforementioned, the occurrence of disinformation is largely due to deliberate misuse by users. Without imposing responsibilities on the user side, it is difficult to regulate the harmful use of AI by users. Meanwhile, it is argued that the logic of AIA is to work backward from certain harms to measures that mitigate the risk that these harms materialize.^[27] The primary focus ought to shift towards the liability associated with the quality of input data, rather than imposing unattainable obligations on data quality.

Apart from the AIA, the Digital Service Act (DSA) aims to govern disinformation. However, the DSA’s legislators only focus on the responsibilities of the intermediary, overlooking the source of the disinformation. Imposing obligations only on intermediaries when LLMs are embedded in services is insufficient, as such regulation cannot reach the underlying developers of LLMs. Similarly, the Digital Markets Act (DMA) focuses on the regulation of gatekeepers, aiming to establish a fair and competitive market. Although scholars recently claim that the DMA has significant implications for AI regulation,^[28] the DMA primarily targets the effects of AI on market structure; it can only provide limited help on LLMs. The problem that the DSA and DMA will face is that both only govern the platform, not the usage, performance, and output of AI per se. This regulatory approach is a consequence of the current platform-as-a-service (PaaS) business model. However, once the business model shifts to AI model-as-a-service (MaaS),^[29] this regulatory framework is likely to become nugatory, as the platform does not fully control the processing logic and output of the algorithmic model.

Therefore, it is necessary to urgently reconsider the regulation of general-purpose LLMs.^[30] The parroting and hallucination issues show that minimal transparency obligations are insufficient, since LLMs often lull users into misplaced trust. When using LLMs, users should be acutely aware that the answers are made-up, may be unreliable, and require verification. LLMs should be obliged to remind and guide users on content verification. Particularly when prompted with sensitive topics, such as medical or legal inquiries, LLMs should refuse to answer, instead directing users to authoritative sources with traceable context. The suitable scope for such filter and notice obligations warrants further discussion from legal, ethical and technical standpoints.

Furthermore, legislators should reassess the risk-based AI taxonomy in the AIA. The above discussion suggests that the effective regulation of LLMs needs to ensure their trustworthiness, taking into account the reliability, explainability and traceability of generated information, rather than solely focusing on transparency. Meanwhile, end-users, developers and deployers’ roles should all be considered in AI regulations, while shifting focus from PaaS to AI MaaS.

---

[1] Take Neeva for example, which is a competitor to which the complaint alludes to at the start. It’s a subscription-based general search engine whose features received mixed comments at this article from The Verge. Some criticized that the only innovation was the model of revenues i.e. that users would rather pay a subscription to avoid ads. Others simply found that Neeva’s search engine was not so much better than Google’s, and that despite shortcomings in search engine optimization, google search was still “good enough.”

[2] Christopher Schell, Yan Zhang and Derek Walters, The Structured Products Law Review, Fourth Edition, Nov 2022, at 62 .

[3] Pittaro, M. L. (2007). Cyber stalking: An Analysis of Online Harassment and Intimidation. International Journal of Cyber Criminology, 1(2), 180–197. https://doi.org/10.5281/zenodo.18794

[4] Article 34 Council of Europe Convention on preventing and combating violence against women and domestic violence (‘Istanbul Convention’), Council of Europe Treaty Series No. 210.

[5] Vidal Verástegui, J., Romanosky, S., Blumenthal, M. S., Brothers, A., Adamson, D. M., Ligor, D. C., … & Schirmer, P. (2023). Cyberstalking: A Growing Challenge for the US Legal System.

[6] https://transparency.fb.com/policies/community-standards/bullying-harassment/

[7] Regulation (EU) 2022/2065 of the European Parliament and of the Council of 19 October 2022 on a Single Market For Digital Services and amending Directive 2000/31/EC (Digital Services Act) OJ L 277, 27.10.2022, p. 1–102.

[8] Article 1(1) DSA.

[9] Directive 2000/31/EC of the European Parliament and of the Council of 8 June 2000 on certain legal aspects of information society services, in particular electronic commerce, in the Internal Market (‘Directive on electronic commerce’) OJ L 178, 17.7.2000, p. 1–16.

[10] Read further on the prohibition of general monitoring obligations: Senftleben, Martin and Angelopoulos, Christina, The Odyssey of the Prohibition on General Monitoring Obligations on the Way to the Digital Services Act: Between Article 15 of the E-Commerce Directive and Article 17 of the Directive on Copyright in the Digital Single Market, Amsterdam/Cambridge, October 2020, https://ssrn.com/abstract=3717022

[11] Recital 41 DSA.

[12] Recital 12 DSA.

[13] Recital 46 e-commerce Directive.

[14] Article 34 DSA.

[15] Art. 15 DSA.

[16] Art. 24 DSA.

[17] K. Baer, R. de Mooji, S. Hebous, M. Keen (2023). Taxing Cryptocurrencies, IMF WP/23/144.

[18] Ibid.

[19] The work is adapted and developed from the preprint version of a paper published in Nature Machine Intelligence, “Zihao Li, ‘Why the European AI Act Transparency Obligation Is Insufficient’ [2023] Nature Machine Intelligence. https://doi.org/10.1038/s42256-023-00672-y”

[20] ‘Much to Discuss in AI Ethics’ (2022) 4 Nature Machine Intelligence 1055.

[21] Zihao Li, ‘Why the European AI Act Transparency Obligation Is Insufficient’ [2023] Nature Machine Intelligence.

[22] Ziwei Ji and others, ‘Survey of Hallucination in Natural Language Generation’ [2022] ACM Computing Surveys 3571730.

[23] Emily M Bender and others, ‘On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?’, Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency (ACM 2021) <https://dl.acm.org/doi/10.1145/3442188.3445922&gt; accessed 14 January 2023.

[24] Marvin van Bekkum and Frederik Zuiderveen Borgesius, ‘Using Sensitive Data to Prevent Discrimination by Artificial Intelligence: Does the GDPR Need a New Exception?’ (2023) 48 Computer Law & Security Review 105770.

[25] Zihao Li, ‘Affinity-Based Algorithmic Pricing: A Dilemma for EU Data Protection Law’ (2022) 46 Computer Law & Security Review 1.

[26] Lilian Edwards, ‘The EU AI Act: A Summary of Its Significance and Scope’ (Ada Lovelace Institute 2022) <https://www.adalovelaceinstitute.org/wp-content/uploads/2022/04/Expert-explainer-The-EU-AI-Act-11-April-2022.pdf&gt; accessed 17 January 2023.

[27] Martin Kretschmer and others, ‘The Risks of Risk-Based AI Regulation: Taking Liability Seriously’.

[28] Philipp Hacker, Johann Cordes and Janina Rochon, ‘Regulating Gatekeeper AI and Data: Transparency, Access, and Fairness under the DMA, the GDPR, and Beyond’ [2022] SSRN Electronic Journal <https://www.ssrn.com/abstract=4316944&gt; accessed 8 January 2023.

[29] Tianxiang Sun and others, ‘Black-Box Tuning for Language-Model-as-a-Service’, Proceedings of the 39th International Conference on Machine Learning (PMLR 2022) <https://proceedings.mlr.press/v162/sun22e.html&gt; accessed 10 February 2023.

[30] Philipp Hacker, Andreas Engel and Theresa List, ‘Understanding and Regulating ChatGPT, and Other Large Generative AI Models: With input from ChatGPT’ [2023] Verfassungsblog <https://verfassungsblog.de/chatgpt/&gt; accessed 20 May 2023.

Copyright © 2023 contributors. This issue and the previous issues of the Transatlantic Antitrust and IPR Developments can be accessed via its webpage on the Transatlantic Technology Law Forum website