Conclusion and Recommendations “ Enactment of limited government regulation would not relieve Silicon Valley of the need to take action of its own. Because of the First Amendment, the U.S. government cannot seek to cleanse online platforms of the many forms of expression that, while not illegal, are still harmful. ” In 2017, when the NYU Stern Center began publishing reports about social media’s effects on democracy, we stressed a preference for industry selfregulation as a way to avoid the First Amendment complications that would accompany government intervention. Those free speech complications remain: Government must not set content policy or participate in substantive content decisions. But some time ago, we realized that the continuing insufficiency of self-regulation meant that government would have to step in, cautiously. In the recommendations that follow, we discuss as an initial matter the responsibilities of YouTube and Google —and by extension, of all of the major platforms. These obligations include far more transparency about how their technology works and how it interacts with human decisions that determine what content users see, which items get amplified, which get down-ranked or removed, and why. We also reiterate a proposal we published in early 2022 for legislation that would enhance the Federal Trade Commission’s consumer protection authority to oversee social media companies. We believe lawmakers and regulators can put this idea into practice without violating the First Amendment’s wise constraint on government interference with speech. Enactment of limited government regulation would not relieve Silicon Valley of the need to take action of its own. Because of the First Amendment, the U.S. government cannot seek to cleanse online platforms of the many forms of expression that, while not illegal, are still harmful. Hateful speech that doesn’t directly incite imminent violence comes to mind, as does most mis- and disinformation, even on vital topics such as public health requirements in the midst of a lethal pandemic. Government can mandate disclosure of data relevant to such harms, and it can require that companies allocate resources necessary to fulfill promises they make about mitigating them. But as a practical matter, the platforms themselves inevitably will bear the bulk of the responsibility for policing expression they host. The scale and speed of their operations, their immediate access to content on their sites, and the complexity of their technology dictate that the platforms cannot be overseen the way a government inspector scrutinizes a meatpacking or pharmaceutical plant. Regulation of social media, by necessity, will be a hybrid affair. HOW YOUTUBE SPREADS HARMFUL CONTENT – AND WHAT CAN BE DONE ABOUT IT 19