Microsoft Promises to Cover Copyright Legal Risk for Some AI Service Customers

Report from the Verge

In Brief – Microsoft has announced that it will back commercial users of its AI-enabled Copilot service if they face legal threats from copyright holders. In a blog post, company President Brad Smith says that Microsoft is making what it calls the Copilot Copyright Commitment because it wants to stand behind customers, understands the concern of copyright holders, and has built guardrails into the service to protect against the possibility of results infringing on copyrighted material. “If a third party sues a commercial customer for copyright infringement for using Microsoft’s Copilots or the output they generate, we will defend the customer and pay the amount of any adverse judgments or settlements that result from the lawsuit, as long as the customer used the guardrails and content filters,” Smith wrote.

Context – There are several copyright policy issues surrounding Artificial Intelligence, especially Generative AI systems built on Large Language Models. At issue here are legal issues surrounding copyrighted works being used to “train” neural networks that then create new works. Some critics claim that the AI services will simply spit out copies of what they digest in training. Microsoft clearly does not see that as a major risk. Generative AI tools are nothing like traditional databases or internet search engines. They don’t store and return fixed data. Instead, they compile realistic, human-sounding responses to queries 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 often pulled from all over the internet. While they often deliver accurate facts and credible responses, we’ve all heard plenty about chatbot “hallucinations”. Rather than copy and regurgitate, they also can simply make things up. A separate issue is the claim by some copyright owners that using copyright works to train AI is itself a copyright violation. More and more are filing suits and demanding payment. Many experts believe that using copyright works in training is likely to be considered “fair use” and therefore legal under US copyright law, although courts will decide.

Meta’s Threads Blocking Controversial Topics in New Threads’ Search

Report from the Washington Post

In Brief – Threads, the text-based social media platform released by Meta in July and linked to its popular Instagram service, has finally added a search function to identify topical Threads posts. However, it quickly became apparent that the service blocked searches on a range of “potentially sensitive” topics, including some that are substantive but controversial, and some that violate platform rules. Although Meta has not released a list of blocked terms, they include “covid,” “long covid,” “coronavirus,” “vaccines,” “vaccination,” “sex,” “nude,” “gore,” and “porn.” When a user searches for a term being blocked, they see a screen with no search results. In the case of COVID and vaccine-related searches, they also see a pop-up linking to the website of the Centers for Disease Control and Prevention.

Context – Threads’ policy to block searches on a range of sensitive topics is itself being called “controversial” by social media experts who argue that social media is an important news and information platform. And some critics are claiming that absent controversial, substantive issues, Threads won’t evolve into the direct competitor to X (formerly Twitter) that many call for. However, from its earliest days, Threads’ leaders have been clear that they did not want the platform going in that direction. In July, Instagram CEO Adam Mosseri said that Threads is “not going to do anything to encourage” politics and “hard news,” and that “the goal isn’t to replace Twitter.” Maybe Meta’s leadership really does see “hard news” as a net negative for its platforms. Look to the current standoff in Canada where the government is trying to force Facebook and Instagram (and Google) to pay government-set licensing fees to media companies when they post their news the platforms. Meta argues that their users don’t value news and the platforms are moving toward entertainment-style content. They are increasingly straightforward about blocking news whenever forced to pay media companies a government-set rate. Canada is clearly a testbed. And arguing that any user thinks of Threads as their go-to place for hard news is ridiculous.

Google Settles With State AGs on App Store Antitrust Case Leaving Epic to Battle Alone

Report from the Wall Street Journal

In Brief – Google and a coalition of 37 State Attorneys General have announced a settlement agreement to end the States’ antitrust complaint against Google for alleged anticompetitive Play Store policies. The AGs’ 2021 complaint has been proceeding in tandem with similar Play Store-focused challenges, most notably from giant app developers Epic Games and Match. The terms of the settlement have not been announced. The parties will have a hearing in front of Federal District Judge James Donato, who has been overseeing the combined cases, on October 12. If Donato accepts the terms, the AGs will withdraw leaving Epic and Match to proceed to a jury trial in front of the judge on November 6. Tim Sweeney, Epic’s CEO and vocal critic of Google’s and Apple’s app store policies, expressed concern that the AG-Google deal was not likely to substantively reduce Google fee rates, a focus of Epic Games in their legal battle with the two mobile ecosystem giants.

Context – This week’s opening of the antitrust trial pitting the US Department of Justice against Google focusing on its search service kicks off a series of big Google antitrust trials that have been brewing for years. The Epic Games v Google battle was started by Epic back in August 2020! Their trial is in November, while Epic’s parallel Apple complaint went to trial in May 2021, saw a decision that August, and already has gone through an appeal in the Federal Ninth Circuit. Apple largely prevailed, except on steering charges. The appeals court opinion criticized Epic’s challenge on Apple’s right to charge commissions. The fact that Epic v Google will be a jury trial is a twist in the recent Big Tech antitrust cases. While Google’s Android is clearly more open than Apple iOS’s walled garden, its brand rules are complicated. Who knows how a jury will react. Then there are the two big antitrust challenges to Google’s adtech business, which may be the most challenging to the company. The DoJ-led complaint may be front of a jury as early as March 2024 while a Texas-led AG complaint filed in 2020 is caught up in a venue fight.

Senators Hold Closed Door Mega-Meeting on Artificial Intelligence and Regulation

Report from NBC News

In Brief – Senators, CEOs from some of the most influential digital technology companies in America, and leaders from top labor, civil rights, and entertainment industry organizations met to discuss Artificial Intelligence and the myriad issues surrounding possible government involvement in its evolution, use, and regulation. The meeting was organized by Senate Majority Leader Chuck Schumer (D-NY) who is leading a bipartisan group of four Senators aiming to quickly educate the body on AI technology. The plan was for two three-hour sessions surrounding a one-hour lunch break, starting with opening statements from the tech industry leaders followed by a discussion moderated by Mr. Schumer. Invitees included CEO-level representatives from Google, Meta, Microsoft, Tesla, Nvidia, IBM, OpenAI, and Palantir, along with leaders from the AFL-CIO, American Federation of Teachers, the Motion Picture Association, Writers Guild, UnidosUS, and the Leadership Conference on Civil & Human Rights.

Context – Is the US Senate remotely close to enacting major AI regulation? No. Not close. ChatGPT’s release may have kicked talk of existential AI threats into high gear, but meaningful legislation and regulation is not measurably closer today than a year ago. The EU, nearly five years into its AI Act plan to create a comprehensive risk-based regulatory model, might finalize something this year. But the EU Parliament’s decision to regulate “foundational models” rather than just real word applications is stoking concerns with over-regulation even there, including from French President Macron. More of the Biden Administration approach marked by principles and corporate commitments to “responsible” AI practices like digital watermark standards is a realistic expectation. If the US Congress legislates, it will be very targeted, such as some plans for federal agencies to deal with AI. Even defining AI would be something. Champions of digital regulation often say the EU is leading, but mid-summer talks at the G7 showed that most governments were focused on AI principles rather than regulation.

Federal Court of Appeals Agrees That Biden Administration Coerced Social Media Platforms

Report from the Washington Post

In Brief – A three-judge panel of the US Fifth Circuit Court of Appeals has ruled that the Biden White House, top government health officials, and the FBI likely violated the First Amendment by coercing or significantly encouraging large social media platforms to restrict posts regarding the coronavirus pandemic and elections. Although the judges upheld that key District Court finding and imposed an injunction on those agencies, the Appeals Court decision significantly narrowed the initial ruling and injunction, striking nine of the ten behavioral “prohibitions” and removing several of the agencies, including the departments of State, Homeland Security, and Health and Human Services, as well as the Census Bureau and Cybersecurity and Infrastructure Security Agency. The one prohibition imposed by the appeals court now reads: “Defendants, and their employees and agents, shall take no actions, formal or informal, directly or indirectly, to coerce or significantly encourage social-media companies to remove, delete, suppress, or reduce, including through altering their algorithms, posted social-media content containing protected free speech.” It applies to the White House, the surgeon general’s office, the CDC, and FBI.

Context – This complaint, brought by the Republican Attorneys General of Missouri and Louisiana, embodies one thread of the Republican social media “censorship” argument, testing the boundaries of the principle that the First Amendment, which prohibits government censorship, also prohibits government coercing private parties to engage in activities the government cannot do themselves. The appeals court ruling appears to strike a sensible balance on that score. The second thread considers the largest digital platforms to be 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 in 2021 and challenged in federal court are top examples. They may be in front of the Supreme Court next year. On those cases, the Biden Administration Solicitor General is encouraging the High Court to protect the right of platforms to moderate content.

Another Amazon Price Parity Policy to Face Consumer Class Action Lawsuit

Report from Reuters

In Brief – Amazon has again failed to dismiss a consumer class action lawsuit targeting practices critics argue force up the price of products sold on the internet. Federal Judge John Chun in the Western District of Washington has ruled that a group of consumers that shop on Amazon can proceed with their lawsuit targeting the Amazon practice called Minimum Margin Agreements (MMAs). MMAs are agreements that Amazon establishes with its retail division’s wholesale suppliers to guarantee that Amazon earns a minimum profit margin even if Amazon lowers its own retail price to match an online competitor. With an MMA, if Amazon price matches a competitor and its own profit margin falls below the guaranteed level, the wholesale supplier of the product is required to pay Amazon the difference. The complaint alleges that Amazon’s wholesale suppliers often respond to MMAs by ensuring that the goods they sell to Amazon are not sold online by other retailers they do business with at a lower price than the price on Amazon to avoid having to pay Amazon when Amazon lowers its price.

Context – The charge that Amazon is engaged in anticompetitive “price fixing” is gaining traction in US courts. Most critics have focused on Amazon’s “price parity” or “MFN” practices that penalize third-party retailers who sell on the dominant Amazon marketplace if they offer lower prices on any other website. While Amazon claims they are only interested in low prices on their platform, many argue that Amazon uses a range of policies to shield its high fees from online competition. Many sellers argue they could sell products at lower prices on other venues that charge lower fees, but if they do so they will be severely penalized on Amazon, which is the largest online marketplace by far. Along with the price fixing lawsuit targeting MMAs, a parallel lawsuit in federal court focusing on third-party sellers has also survived a motion to dismiss. A similar complaint from California AG’s also recently survived Amazon motions to dismiss in California state court, and the issue is likely to be part of a major Federal Trade Commission antitrust complaint expected in the coming weeks.

UK Government Steps Back on Encryption Threat from Online Safety Bill

Report from the Financial Times

In Brief – During consideration of the Online Safety Bill (OSB) in the House of Lords, the UK Government made a statement that appears to commit to not require secure messaging apps and services to have the capability to scan private messages for illegal content. Lord Parkinson of Whitley Bay, who serves as the Parliamentary Under Secretary of State of the Department for Culture, Media and Sport, said that Ofcom, the regulator that will enforce the legislation’s rules, could only require companies to scan their networks for illegal content, such as child sexual abuse material, if it’s “technically feasible,” adding that if technology “does not exist which meets these requirements, Ofcom cannot require its use.” Many security experts argue such technology does not currently exist.

Context – The OSB journey began back in 2017 with a promise by the UK Government to make the country “the safest place in the world to be online”. Over the years, the plans were expanded from the starting point of combatting child sexual material and terrorism advocacy to cover more and more types of bad content, culminating with mandates to combat “legal but objectionable” material. Under fire for threatening free speech, the OSB was trimmed back last fall and refocused, especially on harms to younger users. Lately, the most contentious debate centered around the possible threat to end-to-end encryption, with top messaging apps including Signal and WhatsApp threatening to withdraw their services from the country. The apparent government climbdown was applauded by Meredith Whittaker, the president of Signal, who said “I would call this a victory, not a defeat. And am grateful to the UK government for making their stand clear.” However, Andy Yen, CEO of VPN and messaging company Proton, was more circumspect, saying “all it takes is for a future government to change its mind and we’re right back where we started.” In the meantime, the EU enacted and has begun implementing the Digital Services Act, its version of digital platform content moderation regulation.

The EU Digital Markets Act Covers Five US Companies and China’s ByteDance

Report from CNBC

In Brief – The European Commission has designated six tech giants as “gatekeepers” under the landmark Digital Markets Act (DMA) that regulates the largest digital platforms by imposing 18 duties on them that were previously enforced through antitrust investigations. The six companies are Amazon, Apple, ByteDance, Google, Meta, and Microsoft. They now have six months to bring their most popular platform services into compliance. The law applies to eight kinds of digital services, including social networks, ads, browsers, operating systems, and “intermediation” platforms, which includes platforms like Google Maps. All a gatekeeper’s digital services are not automatically covered by the DMA, but instead each service is evaluated based on the legislation’s criteria. The announcement reveals that Google has the most, with eight covered services, while ByteDance has just one, TikTok. The Commission also opened investigations into Microsoft’s Bing, Edge, and the Microsoft Advertising Platform, and Apple’s iMessage service and iPadOS operating system, to further study if any qualify.

Context – One big question since the DMA was proposed in 2020 has been which companies would be covered. Some argued the measure targeted US companies, including a few Biden Administration officials. European leaders argued it applied to all companies equally, including EU-based businesses. Outside analysts identified approximately a dozen likely companies, including EU-based Booking .com. The EU regulators arrived at a smaller number and no EU-based platforms. The list looks a lot like the recommendation from EU Parliamentarian Andreas Schwab of Germany who shepherded the measure through that body. In 2021 he wanted to limit the bill’s scope to the US giants, saying “Let’s focus first on the biggest problems, on the biggest bottlenecks. Let’s go down the line — one, two, three, four, five — and maybe six with Alibaba. But let’s not start with number 7 to include a European gatekeeper just to please Biden.” Well, two years later, TikTok’s in, Alibaba’s out, and to paraphrase Captain Louis Renault, I’m shocked, shocked, there are no European companies on the list.

Federal Judge Imposes Temporary Injunction Blocking Arkansas Social Media Age Law

Report from the Washington Post

In Brief – A federal judge has temporarily blocked Arkansas from enforcing its new law that requires teenagers under age 18 to have parental consent to create accounts on social media platforms such as Facebook, Instagram, TikTok, and Snapchat. Arkansas’s law is similar to measures enacted in other Republican-led states this year, but it has the earliest effective data. NetChoice, a trade association of large digital platforms, has sued to block the state laws arguing that they are unconstitutional limits on free speech. In his opinion, Judge Timothy Brooks expressed deep reservations about the law’s constitutionality, as well as its efficacy. The Arkansas legislature amended the bill to cover platforms with revenues above $100 million who’s “predominant” function was social media, likely exempting popular services such as Google-owned YouTube and Microsoft-owned LinkedIn, as well online video games, email and messenger services, streaming sites, and e-commerce platforms. Judge Brooks called out that imprecise definition in his opinion, noting that “Act 689 does not explain how platforms are to determine which function is ‘predominant,’ leaving those services to guess whether they are regulated.”

Context – “Protecting” teenagers from social media seems to be good politics everywhere. And, of course, from porn too. Yes, Republican-led states enacted laws requiring parental approval to sign up for social media, but more progressive jurisdictions such as California, France, and the UK are also pushing to create a separate, more regulated version of the internet for teenagers. If there is a “splinternet” coming to Western countries, this is likely what it looks like. 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 causes most privacy advocates to be very worried. Many progressive defenders of at-risk youth warn of their inability to access online services and support if social media is age-gated. Finally, actual studies of teen use of social media show it is not generally harmful and is often a positive force.

New York City Begins Enforcement of Strict Regulation of Short-Term Rentals

Report from the New York Times

In Brief – Following the early August dismissal of an AirBNB lawsuit to block implementation of new regulations in New York City governing short-term rentals, rules are going into effect that may result in tens of thousands of properties being pulled off digital rental platforms by year’s end. The regulations, which have been called a “de facto ban” by AirBNB in their largest US city market, are the latest and potentially most impactful development in a city-by-city battle underway for years. On one side are the platforms and their “hosts”, which include both entrepreneurs who operate networks of rentable properties as full-scale businesses, as well as individual homeowners who rent out rooms, or even sometime their whole home, to help offset housing costs. They argue the rentals create economic opportunity, spread tourism across cities more widely, and allow people to afford their homes. The often influential local hotel industry and allies in local government are on the other side, arguing that rentals are often just unlicensed hotels, highlighting residential neighborhood complaints over transient stays, and claiming short-term rentals are contributing to residential housing shortages and driving up prices.

Context – Short-term rentals have also been highly controversial in top European tourist destinations where similar alliances are engaged. The European Commission has attempted to balance efforts by hotspots like Paris, Amsterdam, and Barcelona to aggressively limit the platform-based hotel alternatives with more welcoming smaller cities and rural communities. The Digital Services Act, now coming online, requires short-term rental platforms to generally support enforcement of local regulations, while new short-term rental platform-specific regulations require data-sharing with governments. However, localities retain flexibility. For example, in Paris, only main residences can be freely rented and with a limit of 120 days per year. Barcelona takes a very different tack, licensing “whole apartment” tourist rentals and instead banning residents from renting spare rooms in their own homes for short stays, the only major city banning that practice.

Top Issues to Follow for Fall 2023 – Best of the Rest

Yesterday, we provided our annual End of the Summer review of the five digital policy questions and issues we think are most timely, important, and interesting for the rest of the year. Today we give a snippet on eight others deserving of consideration.

Implementation of the EU Digital Services Act

Big picture, the most impactful global development continues to be the EU’s relentless drive to establish a new digital platform governance framework with wide-ranging rules and mandates enforced by regulators. It just grinds on. The Digital Services Act is now online with rules for how large platforms handle a wide range of illegal and objectional online content, as well as prohibiting certain advertising practices, and requiring the sharing of data with researchers and authorities. The Digital Markets Act regulating very large “digital gatekeepers” to prevent alleged anticompetitive behavior comes fully into force next spring. The intended level of government oversight reminds many of bank regulation.

Finally Wrapping Up the Microsoft-Activision Deal

Facing Big Tech skepticism from antitrust authorities in the US, EU, and UK, it was always going to be a challenge for Microsoft and Activision to win approval of the $69 billion deal announced in early 2022. The UK CMA rejected the tie-up this April but found itself uncomfortably isolated after the European Commission approved the deal in May based on Microsoft’s market access conditions and a US Federal Judge ruled against Federal Trade Commission opposition in July. The CMA is now reviewing a “substantially different” deal proposal from the companies based on selling Ubisoft Entertainment the cloud-streaming rights for Activision games. Expect approval by their new deadline of October 18.

FTC Antitrust Suit Targeting Amazon Widely Expected

A major antitrust complaint targeting Amazon’s core third-party marketplace and logistics business is expected from the Federal Trade Commission this fall. Ironically, FTC Chair Lina Khan’s law school focus on Amazon may have put them behind the curve. Back then, Amazon was seen as a low-price online retailer trading profits for market share. Today, their first-party retail is a low-margin shrinking business. Smart thinking focuses on Amazon pushing small third-party sellers into the massive Fulfilment By Amazon logistics service that generates high fees and high margins. Amazon has reached an antitrust settlement with the EU on the issue and proposed one to the UK CMA. Will Amazon make a similar offer to the FTC or fight to treat US online sellers and consumers differently than those in Europe?

EU Telecoms and Big Tech “Fair Share” Payment Plans

Telecom companies have long accused large internet platforms of “free riding” on their networks. In Europe, they are aggressively pressing a sender-pays “fair share” plan imposing a new fee on platforms to help fund 5G and broadband. Thierry Breton, Europe’s Internal Market Commissioner sympathizes with charging video streamers such as Amazon, Apple, Google, Meta, Netflix, and TikTok. But critics question whether there is a network investment shortfall, see new fees being passed to consumers, and fear undermining net neutrality. A formal consultation was opened by the Commission in February and opposition is growing among member states, EU Parliamentarians, internet network infrastructure firms, consumer advocates, and industry regulators. But the network operators are not about to turn off their campaign