Archive – 15 Sep 2023

Amazon Aims to Settle Broad UK Antitrust Complaint Along EU Model

Report from TechCrunch

In Brief – Amazon has proposed a settlement offer to the UK Competition and Markets Authority (CMA) to address the antitrust regulator’s investigation opened in 2022 over how the ecommerce and logistics giant determines the products that are most highly recommended to its online shoppers, as well as how Amazon may use data related to the products of third-party sellers to bolster its own retail business. The CMA is inclined to accept Amazon’s offer to treat all sellers equally in its determination to feature a product offering in its “Buy Box”. That promise from Amazon, which will require oversight from an independent trustee, is its attempt to address charges that it has uses the Buy Box algorithm to preference sellers who use its FBA logistics business, pushing sellers to buy the company’s logistics services that often offers fast shipping times but also carries high fees for Amazon. The CMA will be taking feedback from market participants until September 1st.

Context – The CMA’s investigation of Amazon parallels one the European Commission. While both included the long-running concern that Amazon uses data from third-party sellers to unfairly compete with them as a first-party retailer, it is increasingly understood that Amazon’s most profitable core ecommerce business involves third-party sellers who purchase Amazon’s logistics services. That combination of marketplace and logistics services gives Amazon very high commission rates, often reaching 50%, while also maximizing Amazon’s control of the customer experience. Antitrust regulators are now raising issues with how Amazon links selling success on its marketplace to buying the company’s logistics services. In the EU, Amazon’s antitrust settlement last December included agreeing to create a second Buy Box offer with different shipping (presumably slower and less expensive). The offer to the CMA does not include that specific commitment. In addition, reports continue to circulate that a very similar antitrust complaint against Amazon, focused on the Buy Box-FBA link and Amazon setting anti-consumer price floors off Amazon, is likely to come from the US FTC soon.

France Adds to Antitrust Probes of Apple’s Tracking Policy for Third Party Apps

Report from Bloomberg

In Brief – After years of review, France’s antitrust agency has opened a formal probe into Apple’s highly publicized 2021 App Tracking Transparency (ATT) policy that requires third-party apps to get explicit approval from iPhone users to collect online browsing data for use in digital advertising. The ATT policy, initially announced by Apple in 2020, spurred concerned reactions from a wide range of digital advertising industry stakeholders who claimed Apple could use other user data advantages on the iPhone ecosystem to benefit its nascent digital ad business. French trade associations representing internet advertising agencies, technical intermediaries, publishers and others soon filed an antitrust complaint against Apple asking the regulator to block the ATT rollout. The Autorité de la Concurrence rejected that request in March 2021 but announced that it would study Apple’s policy to determine if it was being used to unfairly penalize third-party apps and digital advertisers. The agency has now objected, accusing Apple of “abusing its dominant position by implementing discriminatory, non-objective and non-transparent conditions for the use of user data for advertising purposes.”

Context – Targeted advertising continues to create some of the most interesting policy crosscurrents in the digital policy space. It is more efficient and effective than less-targeted forms of advertising, which benefits small businesses and others with small advertising budgets. In addition, value created by more effective advertising has funded many popular and helpful online services, and users generally prefer “free” ad-based services rather than paying subscriptions. But privacy advocates have long-opposed online data collection and Apple has been at the forefront of harshly criticizing ad-based business models. Although Apple’s mantle of respecting user privacy helped them roll out ATT without government interference, regulators are now engaging, with France joining Germany, Italy, and Poland. The strong growth of Apple’s own ad businesses since 2021 is helping fuel the investigations.

House Judiciary Committee Passes Bipartisan “Fourth Amendment Is Not for Sale Act”

Report from CyberScoop

In Brief – The House Judiciary Committee, which is often the site of highly partisan battles, has passed the “Fourth Amendment Is Not For Sale Act” on a strongly bipartisan vote of 30 in favor and no votes against. The measure prohibits federal law enforcement and intelligence agencies from buying or otherwise acquiring sensitive personal information, including mobile phone location data, from third parties such as data brokers if they would otherwise require a court order to gather the data themselves. The bill also blocks the agencies from acquiring data that was “illegitimately obtained” via an act of deception, hacking or breach of contract, a restriction targeted at services like controversial facial recognition firm Clearview AI that built its service through billions of photos scraped from social media platforms. H.R. 4639 pits privacy advocates and civil libertarians, who decry federal practices that circumvent legal protections against intrusive government searches, against law enforcement organizations who oppose the bill and argue it would prohibit practices routinely used by police investigators to identity and pursue leads in their efforts to combat violent crime.

Context – This bill was initially introduced in the House and Senate in 2021 when Democrats held majorities in both bodies but it did not advance through a committee in either. It is backed by some of the most progressive Democrats and conservative Republicans in the House and Senate. The Judiciary Committee’s debate can be seen here starting at minute 32. The Judiciary Committee’s Republicans, who have been pursuing a series of highly partisan hearings about what they call the “weaponization” of the federal government, are framing this issue as an example. It appears to be one of the few such measures with strong Democratic backing as well. In addition, Chairman Jim Jordan (R-OH) and Rep. Zoe Lofgren (D-CA) noted that the issue should be included in the upcoming debate over reauthorization of Section 720 of the Foreign Intelligence Surveillance Act.

Another Microsoft Teams Antitrust Complaint Filed in Europe

Report from Reuters

In Brief – Videoconference software provider Alfaview, based in Karlsruhe, Germany, has filed a formal antirust complaint with the European Commission accusing Microsoft of unfair and anticompetitive practices in bundling its Teams videoconferencing and messaging service into the broad Microsoft Office 365 software package. Alfaview’s complaint is very similar to one submitted to the Commission in 2020 by chat-style remote collaboration software service Slack. Recent reports have said that the Commission was likely to soon open a formal antitrust investigation of Microsoft’s practices related to Teams. The digital giant is said be offering to no longer require European customers of Office 365 to have Microsoft Teams automatically installed on their devices and reducing the price of Office 365 when Teams was not included. However, sticking points include Microsoft limiting their change in practice to just EU-based customers, rather than globally, as well the amount of the price differential for the Office 365 suite without the Teams application.

Context – Despite being the second most valuable company in the world, trailing just Apple, Microsoft largely avoided the early years of the antitrust “techlash”. However, the acquisition of Activision Blizzard sparked scrutiny from competition enforcers globally. Besides engaging in a full-throated defense of the deal, Microsoft also pursued a global good behavior campaign aimed at progressive regulators, including acquiescing to unprecedented labor organizing, support for major antitrust reforms and app store regulation, and software license changes coordinated with the European Commission to promote EU-based cloud services providers. Microsoft reached a string of 10-year licensing deals guaranteeing access to top Activision games to a range of console, subscription, and cloud gaming platforms, including, most recently, with chief antagonist Sony. The agreements appear to have won over the Commission, which is looking like the lynchpin of a successful campaign overall. But on Teams and Office 365, the Commission is playing the Bad Cop.

Breton Says TikTok Has Work to Do After DSA “Stress Test”

Report from Bloomberg

In Brief – European Commissioner Thierry Breton has reported that a Commission team had carried out a “stress test” at TikTok’s Dublin offices to evaluate the digital giant’s ability to comply with the EU Digital Services Act (DSA), and that the platform has more work to do. The landmark online content moderation legislation requires digital platforms to be compliant with its new standards by August 25. The most stringent duties are being imposed on “Very Large Online Platforms” (VLOPs) that have at least 45 million monthly users in the EU, including oversight by the European Commission rather than a national regulator, stricter criteria for dealing with objectionable material, and the submission of regular risk assessments. Nineteen digital platforms, including two search engines, were designated as VLOPs, including TikTok. The goal of the stress test was to determine the state of the company’s capabilities to deal with DSA rules on issues including content recommendations, illegal and objectionable content such as child sexual abuse materials, data access, and transparency of platform rules and practices.

Context – The EU is relentlessly putting in place a digital platform governance framework with wide-ranging rules and mandates to be enforced by regulators. Along with the 19 DSA VLOPs, seven of the largest digital platform companies, including ByteDance, have informed the EC that they fit the “gatekeeper” criteria under the Digital Markets Act. Stress tests are regulatory exercises which are most notably used in the financial services sector to try to project how banks and other regulated institutions would handle various economic or business shocks. However, unlike in the long-regulated banking sector, standards and practices for online content moderation by various kinds of platforms are not widely established, so the process is being developed on the fly, including on controversial topics such as free speech and political unrest. Twitter and Meta have also discussed plans to participate in pre-deadline stress tests. On the other hand, two VLOPs, Amazon and German online retailer and marketplace Zalando, have filed suit to be pulled out of the VLOP regime entirely.

Regulating How the US Government Handles AI – Rare Legislative Progress

Report from Politico

In Brief – Despite near daily talk of the need for government to regulate artificial intelligence, concrete legislative progress in the US Congress is rare. One major exception is legislation coming from the Senate Homeland Security Committee focused on the use of AI by the federal government itself. Sen. Gary Peters (D-MI), the committee chair, is leading the effort, but it has bipartisan backing. The Transparent Automated Governance Act, which sets transparency standards for federal agencies using AI and calls for citizen appeals processes for AI-influenced agency decisions, and the AI Leadership Training Act, which establishes new AI training programs for managers and employees at federal agencies, have been passed out of the committee. The recently introduced AI LEAD Act requires federal agencies to appoint a “chief AI officer” to oversee AI system acquisition and usage and creates a “chief AI officers council” to develop a government-wide AI strategy. The policies embodied in the bills were developed before the recent AI hype cycle, and although narrower in scope than many AI regulation advocates call for, Peters argues that if the federal government puts AI rules in place for itself, similar rules will more easily spread to the private sector.

Context – At the other end of the spectrum, the EU seems committed to being the first major market to have a comprehensive AI regulatory regime. However, the carefully crafted AI Act plan focused on high-risk AI applications was dramatically expanded by the EU Parliament after ChatGPT exploded on the scene. Regulating “foundational models” rather than specific applications moves away from the broader global consensus and is making some European businesses and governments nervous. Potential limits on government surveillance is also a divisive issue in the EU. On the other hand, global discussions about limits on AI-enabled weapons systems, such as at the UN, could be a focused starting point for international AI cooperation. Back to the US Congress, bipartisan legislation to prohibit AI from launching US nuclear weapons, a seemingly non-controversial policy, also has been introduced.

Big AI Companies Sign onto White House Generated AI Safety Pledge

Report from the Washington Post

In Brief – Top companies developing generative AI services, including Microsoft, Google, Meta and OpenAI, have pledged to build safeguards into their operations to help address concerns with the potential negative impacts of AI-enabled systems. Concerns include accelerating misinformation, cybercrime, biases in hiring, lending, insurance and other commercial services, threats to jobs, privacy, and the illegal misappropriation of intellectual property of artists, writers, and other creators. With substantive legislation from the US Congress likely a way off, the Biden Administration is rallying companies, starting with the largest, to agree to implement “responsible” AI practices such as helping combat AI-generated fakes with industry standard digital watermarks, allowing outside experts to test AI system security, capabilities, and vulnerabilities before public release, and investing in research on AI’s risks to society. The Administration is also planning to press for similar AI governance frameworks in key markets, including Britain, Germany, Japan, and South Korea.

Context – The public release of OpenAI’s chatbot phenom ChatGPT last fall kicked into high gear talk of existential AI threats and the need for government to regulate the technology. Is meaningful regulation likely soon? In Europe, over a time horizon of a few years, yes. The EU, the Western leader on technology regulation, is nearly five years into the most comprehensive legislative effort to regulate. Its AI Act plan was built on a risk-based regulatory model that focused on policing AI-enabled applications, not the underlying technology. But with ChatGPT’s rollout, the European Parliament’s version added regulation of “foundational models”. Elsewhere, not as soon. If the US Congress legislates, expect narrow measures starting with how the federal government uses AI services. In addition, recent AI policy talks at the G7 point to most of the governments being focused on promoting principles of safe AI development rather than regulation like in the EU and China. And even the European Commission is pressing companies to join a voluntary AI Pact to precede the full AI Act.

UK Government’s Online Safety Bill Tweak Fails to Mollify Encryption Defenders

Report from the BBC

In Brief – In the face of ongoing criticism from secure messaging apps, including Signal, Meta’s WhatsApp, and Apple’s iMessage, the UK Government has made a small change to a controversial section of the Online Safety Bill (OSB). The messaging apps and privacy advocates argue that the OSB will eliminate true end-to-end-encryption (E2EE) that they claim is essential to protecting user privacy by allowing regulator Ofcom to require a messaging platform to have the capability to determine if a message contains illegal content, such as child sexual abuse material. Most security experts argue that E2EE is secure because only a message sender and recipient can unlock a message’s encryption. The companies have said that they will withdraw their messaging services from the market rather than compromise their E2EE regime to give themselves, or any other third party, the ability to break encryption. They have called for the OSB to be amended to explicitly clarify that Ofcom cannot require a platform to have the ability to monitor private communications. The UK Government continues to reject that change, arguing that platforms should be able to offer private communications services that are also free from illegal content, a position backed by child protection advocates. Instead, the OSB was amended in the House of Lords to require Ofcom to receive a report from unspecified “skilled persons” supporting any request to order messaging content to be scanned by a platform, a tweak critics argue falls short of even requiring judicial oversight of Ofcom requests.

Context – The UK Government has said the OSB will make the country “the safest place in the world to be online”, but that regulatory aspiration appears schizophrenic alongside pro-innovation digital policies. The Chief Executive of London-based Element Software recently said that a threat to strong encryption that leads to a pull-out of major messaging apps will have “an incredibly chilling effect on the whole London tech scene.” Put this standoff right there with the UK CMA’s reconsideration of being the last hurdle for the Microsoft-Activision deal.

American Economist Withdraws From Top EU Economist Spot Under French Fire

Report from Politico

In Brief –  Following fierce criticism from numerous French Government and European Parliament leaders, American economist Fiona Scott Morton, a high-profile progressive who served in a top position in US Department of Justice during the Obama Administration, withdrew from her appointment as the chief economist for the European Commission’s competition authority. Critics focused on Scott Morton’s nationality and asked if there wasn’t an equally capable European interested in filling the key role as the Commission continues its efforts to rein in predominantly US-based corporate tech giants. Concerns were also been raised about conflicts of interest created by past work Scott Morton had done on behalf of Amazon, Apple, and Microsoft, three of the digital gatekeepers being scrutinized in Europe that she defended in the past. Scott Morton had been named to the prestigious role by European Commissioner Margrethe Vestager, who leads the Commission’s antitrust efforts and is also a key leader on digital policy, and the commissioner steadfastly defended the hire and the economist from the criticism. In addition, a collection of 39 top European economists publicly backed Scott Morton’s qualifications, experience, and commitment to pushing back on anticompetitive Big Tech practices.

Context – Internecine division over a moderately influential government position has also sprung up in the US, with a small but vocal trio of anti-Big Tech conservative groups calling on Senate Republicans to reject the nomination of Melissa Holyoak, Solicitor General of the State of Utah, to fill one of the two open Republican Commissioner spots at the Federal Trade Commission. The FTC, which is increasingly at the center of partisan fights over a range of antitrust and regulatory issues, has been operating without any Republican Commissioners since the resignation of Christine Wilson in February, which included pointed criticism of Chair Lina Khan and her top staff for repeatedly exceeding the agency’s legal authority and breaking its bipartisan norms.

Meta Challenging European Commission’s Facebook Marketplace Antitrust Complaint

Report from Bloomberg

In Brief – Meta representatives have met with European Commission officials in a closed-door hearing giving the social media giant an opportunity to formally refute the Commission’s statement of objections that the company breached EU antitrust rules by distorting competition in the market for online classified ads to benefit its own Facebook Marketplace service. The company launched the Facebook Marketplace in 2016 and quickly integrated it into its core social media platform, which is one of the most widely used digital platforms in the world. Facebook Marketplace has since grown into one of the top classifieds ads platforms in many markets, including in Europe. The Commission initially flagged two practices as undermining competition, the tying of Facebook Marketplace to the dominant Facebook social network service, and the imposition of unfair terms and conditions, including regarding access to user data, on third-party classified ads and marketplace businesses that advertised on Facebook, in ways that benefited Facebook Marketplace in competing with those services.

Context – Facebook’s ability to grow the FB Marketplace into a classifieds’ giant using data from, and links to, its core platform is an example of platform “self-preferencing” targeted by advocates of more robust Big Tech antitrust regulation. The UK’s Competition and Markets Authority opened a parallel investigation of Facebook and its Marketplace in 2021. Meta has offered to resolve the UK investigation by allowing advertisers on Facebook to opt-out of their data being used by or for the Facebook Marketplace. The CMA has indicated that it is inclined to accept the offer and is gathering UK market feedback. One would expect a similar offer to be made to the European Commission, but the case is also relevant in the context of the EU’s Digital Markets Act, which was enacted last year to address conduct like this by “digital gatekeepers” through ongoing regulatory processes rather than case-by-case competition law enforcement. Expect this type of concern to be shifting to DMA regulatory reviews going forward.

California State Senate Consideration of Media Payments Bill Delayed Until 2024

Report from the Los Angeles Times

In Brief – The California Journalism Competition and Preservation Act, legislation forcing large digital platforms to pay news media companies a “journalism usage fee” when news content they publish appears on the platforms next to advertising, has been designated a “two-year bill” in the California legislature, pushing State Senate consideration into 2024. The bill passed the State Assembly in June by a vote of 55-6. Although initially scheduled for Senate committee consideration in July, Senator Tom Umberg (D-Orange) has said he will hold an informational hearing this fall to further explore issues the bill attempts to address and look at examples of successful legislation in other countries.

Context – Newspapers have been complaining for two decades that the internet ruined their business model. In recent years they started lobbying governments to force digital ad giants Google and Meta to pay them when news appears on their platforms. Australia was an early testbed, enacting legislation in 2021 to compel payments through forced arbitration. But they never pulled the trigger on arbitration talks because both platforms increased payments to Australian media. As other jurisdictions explored similar tactics, both companies drew policy lines in the sand. Google rejects paying for basic search links. Meta rejects paying for news content posted solely by users, including media companies themselves. In addition, both created curated news services and paid hundreds of millions of dollars to media companies. On those “voluntary” payments, the two digital giants appear to be moving apart. Google is expanding theirs. Meta says it is moving away from paying for news content, instead shifting to TikTok-style entertainment content. Attention is now squarely focused on Canada, which recently enacted forced payments legislation. Meta quickly responded that they would drop media content from Facebook and Instagram in Canada rather than pay when others post. More surprisingly, Google also announced that they would drop Canadian media posts from search results rather than pay for basic links.

European Commission Imposes Maximum Fine on Illumina for Closing Grail Deal in 2021

Report from Reuters

In Brief – Genetic testing company Illumina has been fined 432 million euro ($476 million) by the European Commission for closing its acquisition of cancer test maker Grail in 2021 before securing EU antitrust approval. The record fine is the latest exchange in the lengthy high stakes battle between the medical tech companies and competition regulators in Europe and the United States. While the Commission ruled that the acquisition violated its antitrust law, would harm competition, and needed to be unwound, Illumina argued that the EU did not have jurisdiction over the deal in 2021 because the small test maker had no operations or sales in Europe, and because the deal fell below the company revenue threshold for Commission acquisition reviews. However, Illumina has suffered repeated setbacks at the Commission and in EU courts challenging the review.

Context – Microsoft’s $75 billion acquisition of Activision-Blizzard has garnered more media coverage, but Illumina’s battle to reacquire Grail is central to more acquisition policy issues. Yes, the FTC, led by antitrust activist Lina Khan, has suffered another court setback on Microsoft-Activision, with the Ninth Circuit Court of Appeals rejecting the FTC’s appeal of their federal district court loss. Illumina-Grail also involved a big FTC setback challenging a vertical merger, with an FTC administrative judge rejecting the FTC bid to block the deal in 2022. But in that case, the FTC Commissioners voted to block the deal anyhow. The companies are now appealing that decision in federal court, where the agency keeps losing, and where its administrative court processes themselves may be threatened. In Europe, the Commission’s decision to approve the huge Microsoft deal appears to have been the Good Cop that unlocked the giant acquisition, but its stubborn refusal to bend on its authority to review deals involving a US company with no European business, but which might someday, threatens to add to partisan and ideological divisions in the US over the proper role of foreign competition laws and regulators on US antitrust law and business operations.

The Federal Trade Commission is Investigating AI-Phenom OpenAI

Report from the Washington Post

In Brief – The Federal Trade Commission (FTC) has opened an investigation into OpenAI, the maker of AI-enabled chatbot phenom ChatGPT, questioning whether the company has harmed consumers and violated federal laws. The agency sent OpenAI 20 pages of questions and document requests regarding company practices and policies dating back to mid-2020. The investigation covers company data collection and handling practices, data security and concerns with potential data breaches, policies related to users and queries, and practices regarding how its services produce responses regarding individuals, which can include accurate personal information, or inaccurate, even wildly inaccurate, material.

Context – ChatGPT has shaken up discussions of AI regulations everywhere. It is important to get grounded in the fact that AI is not one kind of technology. Digital surveillance, self-driving vehicles, automated weapons systems, the list goes on and on. All different. ChatGPT involves generative AI, large language models, and neural networks. A key learning regarding these tools is that chatbots and other large language models are not traditional databases or internet search engines. They don’t store and return fixed data. Instead, they are built to produce realistic, human-like responses to chat-style queries. They do this by determining which fragments of text best follow other sequences, all based on a statistical model that has ingested and processed billions and billions of examples pulled from all over the internet. It is entirely possible that they deliver factual accuracy, credible ideas, and responses that seem sensible, even creative. But being “correct” is not the primary goal of the tools. They can also produce “hallucinations”, a term used for wildly-inaccurate responses. OpenAI has received similar questions on privacy and incorrect responses from European regulators, including being temporarily banned in Italy but is back online. Google initially withheld Bard, its AI-enabled chatbot, from Europe in the face of similar regulatory concerns, but has now released the service with policies like those negotiated in Italy by OpenAI to restore ChatGPT.

European Commission Wraps Up Latest US-EU Agreement on Data Transfers

Report from the New York Times

In Brief – The European Commission has determined that the EU-US Data Privacy Framework adequately protects the privacy rights of Europeans regarding their personal data held by companies in the United States. The agreement aims to resolve a European legal dispute that has threatened to interrupt the ability for countless companies, including giants like Meta and Google, but also medium-sized and small firms in both jurisdictions, to transfer, store and process data for European-based users in the US. The agreement sets out more clearly when US intelligence agencies can access personal information about people in the EU and outlines how Europeans can appeal such collection. In particular, the Biden Administration has created the Data Protection Review Court, an independent review body made up of American judges, to hear appeals from EU citizens who allege that their personal data was improperly collected and used by US agencies. European privacy advocates have criticized the pact as again falling short of European law and promised to challenge and overturn this effort as well.

Context – We are 10 years into the US-EU legal dispute over “Cross Border Data Flows” following the Snowden revelations. The European Court of Justice invalidated the US-EU Safe Harbor agreement in 2015 and then invalidated the follow-on US-EU Privacy Shield agreement in 2020. Both cases nominally pitted Facebook, which stored and processed data on their EU users in the US, against EU privacy activists, but the fights were about US intelligence and anti-terrorism surveillance laws. The US Congress haven’t changed the laws, and EU leaders have continued reaching deals with the US to avoid major digital service shutdowns and engage in antiterrorism surveillance as well. The same privacy advocates are highly critical of the new deal. So, the most important news out of Europe will be how the European Court of Justice eventually rules on the new deal. Meta, meanwhile, faces a $1.3 billion fine from the Irish data protection authority for operating under the earlier agreements and in line with countless other US and European companies.

Google Hit with Latest Lawsuit Targeting AI Training Copyright and Privacy Policy

Report from CNN

In Brief – Google is the latest big company with a “generative AI” service to be targeted in a class action litigation campaign alleging violations of privacy and copyright law by gathering massive amounts of data on the internet and using it to “train” AI systems. Microsoft, OpenAI, and Meta also face similar federal suits. The plaintiffs are asking the court to halt the company practices and provide people with the ability to opt out of their data being used. A Google spokesperson said, “American law supports using public information to create new beneficial uses, and we look forward to refuting these baseless claims.”

Context – Copyright and privacy are increasingly central to debates surrounding generative AI systems because their “neural networks” process unimaginably large bodies of data, such as most of the texts, images, or music on the internet, to learn how things appear. That often includes huge amounts of both copyrighted material and content created by and about people. But neural networks are not like traditional databases or search engines. They don’t gather, classify, store, or retrieve training data. Instead, they “train” using the existing material and then generate new output in response to queries. Of course, like with human creators, everything created by an AI system may resemble something that already exists and was reviewed. The copyright principle of “fair use”, which is relatively robust in the US legal system, will be central to any legal challenge regarding AI training. One insight from the Supreme Court’s recent Warhol v Goldsmith copyright fair use decision is that it may prove very important whether courts see the purpose of companies using copyrighted works to train AI as the building of a distinct “large language model” service, or as the creation of the service’s individual outputs that may compete in the market with the unnumbered copyrighted works that the system learned from. The European Parliament’s version of the AI Act includes a first step towards regulating generative AI foundation models including with a transparency requirement to identify copyrighted materials used in training.

OECD Tax Meeting Produces Agreement to Extend Digital Tax Truce Through 2024

Report from Reuters

In Brief – Tax and finance officials from 138 member countries in the Organization for Economic Cooperation and Development (OECD) have agreed to extend through the end of 2024 the current freeze on new national digital taxes as part of the now stalled global corporate tax reform effort being shepherded through the organization. Only five countries – Belarus, Canada, Pakistan, Russia, and Sri Lanka – did not support the extension. Canada, which plans to implement a national digital services tax (DST) at the end of 2023 despite strong US opposition, objects to more delays when several European countries are already collecting revenue under DSTs that were in place before the tentative OECD deal was reached in July 2021.

Context – The two-part tax reform deal negotiated at the OECD attempts to address two long-running tax concerns – the right way to tax digital giants like Google and Apple, and frustration that some countries offer “tax havens” to multinational companies to win investment. The first part, dubbed “Pillar 1”, was initially a new tax on approximately 20 very large, mostly US-based, digital companies, but was modified in 2021 to cover 100 or so highly profitable consumer-facing businesses, expanding beyond digital platforms. It benefits countries with large numbers of wealthy consumers by enabling them to increase taxes on consumer goods companies based in other jurisdictions. The second half, “Pillar 2”, aims to undermine tax havens by having countries agree to tax multinational companies at least 15 percent. Progress was made in 2021 when the Biden Administration shifted the US negotiating priority from fighting digital taxes to achieving the global “minimum tax”. However, the overall effort has hit hurdles in the US, Europe, and with some developing countries. Several European countries, as well as India and Turkey, started imposing DSTs on some US companies in 2022. The Biden Administration negotiated agreements for them to rebate tax overages back to the companies once the full global agreement went into effect, but those deals expire in 2023. We’ll see if they are extended as well.

California Judge Delays Enforcement of California Privacy Rights Act Regulations

Report from MediaPost

In Brief – A California State Superior Court Judge has issued a divided ruling that delays until March 29, 2024, enforcement of regulations issued by the California Privacy Protection Agency (CPPA) on March 29, 2023. The regulations implement many of the provisions of Prop. 24, the Consumer Privacy Rights Act (CPRA), enacted by state voters in November of 2020. Prop. 24 called for the CPPA to issue regulations by July 1, 2022, with the regulations becoming effective one year later. The CPPA did not meet this timetable, issuing regulations on 12 of the 15 CPRA sections on March 29, 2023, while three sections remain unfinished. However, the regulator planned to stick to July 1, 2023, for enforcing the rules. The California Chamber of Commerce sued to block enforcement until one year after all CPRA regulations were issued, arguing that the ballot initiative provided companies a full year to develop plans to implement the new requirements. Judge James Arguelles backed the request for a year delay in enforcing the rules issued on March 29, 2023, but did not support delaying all enforcement until the final three sections were also completed. In addition, the CPPA argued that some CPRA rules were already covered by the California Consumer Privacy Act (CCPA) enacted in 2018 and therefore did not require a further year for business to comply, a view also backed by Judge Arguelles.

Context – Florida recently became the tenth state to enact data privacy legislation since California began the trend in 2018. Success with state bills has largely been linked to Democrats pulling back from insisting on class action lawsuits being part of the enforcement mix, which has been a feature of California’s CPPA and CPRA. Similar data privacy laws were enacted in Democrat-controlled states like Virginia, Colorado, and Connecticut in 2021 and 2022. The latest string of wins has been in Republican-led states. While the state laws all differ some, compliance complexities are not boiling over. And the class action tsunami from Illinois’ biometric privacy law is an ongoing warning on untethered class action litigation risks.

EU Kicks Off Deep Dive Review of Amazon-iRobot Vacuum Deal

Report from Reuters

In Brief – The European Commission has opened an in-depth investigation of Amazon’s proposed acquisition of iRobot, the maker of the Roomba robotic vacuum. The ecommerce giant announced the $1.7 billion deal in August 2022. The Commission’s competition authority said that it was concerned that the deal might threaten competition in the market for robot vacuum cleaners, such as by enticing Amazon to tilt search results on its massive marketplace to benefit iRobot products and disadvantage competitors such as Dyson and Shark. The regulator also raised concerns about how Amazon might broadly use iRobot user data to gain advantage over current or future marketplace competitors. The Commission has set a target date of November 15, 2023, to formally approve or reject the deal.

Context – Welcome to another round of Antitrust Enforcement Roulette with the US, EU, and UK regulators. In terms of financial value, Amazon’s bid for iRobot pales in comparison for Microsoft’s attempt to buy Activision-Blizzard, but they are two examples of a lack of consistent treatment of tech deals. Roomba’s market share slide, iRobot’s lack of profitability, and the fact that Amazon did not have its own robot vacuum, all raised questions about the strength of a legal challenge. Yes, there is general opposition to Big Tech acquisitions, and the non-traditional privacy charges regarding in-house Roomba videos and maps have stood out most and are similar to privacy accusations raised about Tesla in-car videos. The UK CMA, who rejected Microsoft-Activision, signed off on Amazon-iRobot and very directly rejected the idea that Amazon had a financial incentive to undermine competitor vacuum sales on its marketplace given the high commissions the ecommerce giant makes on third-party vacuum sales. The European Commission recently signed off on Microsoft-Activision, accepting a remedy package allowing cloud gaming competitors access to Activision games that the CMA rejected. The US FTC rejected the Microsoft deal and asked a federal judge to block it from closing (but was turned down), but has not made its call on Amazon-iRobot.

France Passes Social Media Age Verification and Parental Consent Law

Report from the France 24

In Brief The French legislature has approved a bill requiring social media companies to adopt technical means to verify all users’ ages and obtain parental consent for those under 15 years of age. The measure adds to momentum building in many jurisdictions to increase barriers to younger teens using social media and other digital services, including pornography, in the face of charges that their use is often harmful to mental health. Although social media sites are technically not available to children below age of 13 in France, the French National Commission for Technology and Freedoms (CNIL) claims that more than half of youths aged 10-to-14 use social media sites like Snapchat and Instagram. Although the French Government said that the law would be “applied as soon as possible”, the exact date that it will come into force remains unclear. The European Commission will need to confirm that it conforms with EU law and then social media sites will have a year to comply with the policy for new subscribers, and a further two years to apply the requirements to existing users.

Context – If there is a “splinternet” coming to Western countries, it is most likely going to involve age. US States with undivided Republican control have been especially quick off the mark, with Arkansas and Utah already enacting strict statutes requiring social media platforms to get parental approval for teens under age 18 setting up social media accounts. Porn sites are unsurprisingly also a target, with seven US States imposing age verification. But California, France, and the UK are examples where far more progressive governments are pushing to create a separate, more regulated version of the internet for teen. We don’t have the space here to critique too deeply, but super-smart analyst Mark Masnick does here, here, and here. Creating internet-wide age verification also causes most privacy advocates to be worried. Finally, actual studies of teen use of social media show it is not generally harmful and is often a positive force. In the US, there will be constitutional challenges but the situation in Europe is less clear.

Top Messaging Apps Issue an OSB Encryption Warning to UK House of Lords

Report from Evening Standard

In Brief – Top executives from leading secure messaging apps Whatsapp and Signal, and UK-based Element, have strongly reiterated their objection to the current Online Safety Bill (OSB) that they claim will undermine strong end-to-end-encryption (E2EE) and argue that changes are needed to protect user privacy and security, as well as allow the UK to serve as a tech industry hub. The companies insist that the OSB gives government regulator Ofcom the authority to require a messaging platform to have the capability to scan user messages for illegal content, such as child sexual abuse material (CSAM). But a messaging company using true E2EE cannot monitor messages because only a sender and recipient can unlock a message’s encryption. A wide range of UK security experts argue that the bill will prohibit true end-to-end encryption. Bill backers argue that the legislation does not “ban” strong encryption, but they also insist that messaging platforms must be able to staunch the flow of illegal content threatening young people. WhatsApp and Signal have said that they would withdraw from the UK market rather than weaken their encryption protocols, and the Chief Executive of London-based Element Software said, “It’s going to be an incredibly chilling effect on the whole London tech scene.”

Context – The OSB, which the UK Government has said will make the country “the safest place in the world to be online”, highlights its schizophrenic digital policies and the criticisms from Element Software are likely to resonate. Along with the high-profile encryption battle, the OSB threatens jail time for an expanded range of corporate executives if digital companies fail to police CSAM and calls for online age verification that worries many privacy advocates. The foundation behind Wikipedia is even threatening to withdraw from the market. Although the UK Government pitches an AI strategy that it claims is less regulatory than the EU, its Digital Markets, Competition and Consumers bill largely parallels the EU’s Digital Markets Act, and the UK competition regulator has been particularly aggressive on tech acquisitions, in several cases moreso than the EU.

Federal Judge Rejects FTC Effort to Block Microsoft’s Acquisition of Activision-Blizzard

Report from the Wall Street Journal

In Brief Federal District Court Judge Jacqueline Scott Corley has ruled in favor of Microsoft and Activision-Blizzard and against the US Federal Trade Commission (FTC), rejecting the regulator’s request to issue an injunction blocking the proposed $75 billion acquisition of the giant videogame developer. In her opinion, the judge said that the FTC had failed to show that Microsoft owning Activision games would harm competition in the videogame console or cloud-gaming markets, and instead that the deal would more likely lead to “more consumer access to Call of Duty and other Activision content.” The FTC announced its objection to the deal last December and challenged it in the agency’s special court system with a trial scheduled in August. The regulator filed for an injunction in US District Court to block the acquisition from being completed while the administrative proceedings were underway, a route it increasingly appeared the companies would take. Judge Corley’s decision increases the chances that the deal will be closed soon, even as the companies work to overturn a rejection by the UK antitrust regulator. The FTC may appeal Corley’s decision, as well as continue with its administrative court challenge even if the deal is completed, but the companies appear to hold the legal high ground

Context – The Biden Administration antitrust regulators are very skeptical of acquisitions, especially by Big Tech, but when they challenge tech deals in court, they are not proving very successful. Last year, a similarly placed federal judge turned down an FTC injunction request to block Meta’s acquisition of startup VR app developer Within Unlimited. An administrative judge in the FTC’s own court system turned down their case to block the Illumina-Grail deal. The latest setback affirms the independence of the US judiciary, reminds observers of the strength of the US competition law status quo, and further isolates the UK Competition and Markets Authority, which rejected the deal in April only to see the European Commission approve it in May and now have its ally in the US FTC being undone in US federal court.

Amazon Challenges EU DSA Designation as a Very Large Online Platform

Report from Reuters

In Brief – Amazon, the largest online marketplace in Europe, has filed a challenge in the EU General Court opposing the European Commission’s decision to designate it as a “very large online platform” (VLOP) under the Digital Services Act (DSA). The online content moderation law considers digital platforms with more than 45 million monthly active users to be VLOPs. They face a more stringent regulatory regime, including oversite by the European Commission rather than a national regulator, stricter criteria for dealing with objectionable material, and the submission of regular risk assessments. Nineteen large digital platforms, including two search engines, were designated as VLOPs. Amazon argues that they are not the largest retailer in any country of the EU and that holding them to different standards than those retailers is unfair. The Commission responded to the Amazon complaint noting that “The scope of the DSA is very clear and is defined to cover all platforms that expose their users to content, including the sale of products or services, which can be illegal. For marketplaces as for social networks, very wide user reach increases the risks and the platforms’ responsibilities to address them.”

Context – The EU is relentlessly putting in place a digital platform governance framework with wide-ranging rules and mandates to be enforced by regulators. Along with the 19 DSA VLOPs, seven of the largest digital platform companies have informed the EC that they fit the “gatekeeper” criteria under the Digital Markets Act (DMA). Amazon is on that list as well. Other digital companies are looking to escape from either the DMA or DSA VLOP status as well. Bytedance believes that TikTok should be exempt from some DMA mandates as a “challenger” platform. Zolando, a Germany-based online retailer and marketplace, which happens to be the only EU-based company among the 19 DSA VLOPs, has also filed suit in the EU General Court to be taken off the list, with its CEO saying, “These companies are talked about as bad actors and all of a sudden we are on the same list. This is bad for our brand.”

Judge Issues Injunction to Block Administration Pressure on Social Media Platforms

Report from the Washington Post

In Brief – US District Court Judge Terry Doughty has issued a landmark ruling blocking a wide range of senior federal government officials, including in the Biden White House, from communicating with social media company officials to encourage certain content moderation actions. The injunction, which has been appealed by the US Department of Justice, was requested by the Attorneys General of Missouri and Louisiana as part of their federal lawsuit accusing Biden Administration officials of pressuring Meta, YouTube, Twitter, and others to suppress content on COVID, vaccines, election integrity, and other issues. The case tests the boundaries of the principle that government in the US, bound by First Amendment limits prohibiting censorship, cannot coerce private parties to engage in activities the government cannot do themselves. Having rejected the DoJ’s motion to dismiss the case in March, and now issuing this injunction, Doughty seems inclined to side with the Republican State AGs. The judge’s order does not limit the content moderation authority of the platforms themselves and also sets out a range of exceptions for communications between government officials and the companies, including to warn them of national security threats, criminal activity, or voter suppression.

Context – There are two parallel threads of the Republican social media “censorship” argument. One is that the largest platforms are either public venues limited by the First Amendment directly, or a type of “common carrier” prohibited by law from policing content. The social media laws passed in Texas and Florida are examples. They may be in front of the Supreme Court next year. This suit represents the other. Republicans on the House Judiciary Committee are also pressing it. Federal anti-conservative censorship from when Donald Trump was President (Hunter Biden laptop) make little sense, but Biden officials pushing platforms to police COVID “misinformation” in 2021 is a more credible allegation. Governments in countries as diverse as the EU, UK, Australia, India, and Vietnam are not similarly constrained.

Google Will Block News In Canada Rather Than Pay Media Companies for Search Links

Report from Reuters

In Brief – Following the passage of Canada’s Bill C-18, the Online News Act, Google has announced that it plans to block the appearance of Canadian news content in its search results rather than pay Canadian news outlets when media content available on the internet is reached via Google search. The announcement follows Meta’s similar response to the Canadian law to require the digital giants to pay Canadian media companies. The measure, somewhat modeled after Australian legislation enacted in 2021, goes into effect in six months with the Department of Canadian Heritage to finalize implementing regulations and guidance directing the payment negotiations and binding arbitration process when deals are not reached. During the Australia showdown, Meta and Google tried to draw lines in the sand on media payments, with the social media giant rejecting paying for media content posted by users, including when news media companies post their own content, and Google objecting to paying for basic internet search results and links. Both companies admitted to testing changes to their Canadian services to cut off Canadian media content in the run up to Bill C-18’s passage, and now both have announced their intentions to prohibit Canadian media content if the bill is implemented in its current form.

Context – In recent years, Meta and Google both created curated news services on their platforms as vehicles to pay hundreds of millions of dollars to media companies. The goal was to reduce political heat. Google continues to announce expansions, including plans to open the Google News Showcase in the US and a system to pay licensing fees for the use of news snippets in Europe. Meta has been moving in the other direction on paying news companies. They increasingly claim most social media users prefer TikTok-style entertainment and that their focus will shift there. “News” also brings misinformation, angst, and strife. However, the Canadian law crosses red lines for both companies. Sustained shutdowns in basic platform services in Canada will be big news. And a similar bill has momentum in California.