AI Preemption in Exchange for Unconstitutional Online Safety Bills is a Bad Deal
Various press reports have suggested that the White House is considering a grand compromise with major tech firms: in exchange for throwing their support behind a package of controversial online safety bills, the firms will win some federal preemption of state and local artificial intelligence (AI) regulations.
It is a bad deal and should be rejected by the companies, the public, and policymakers.
While many congressional lawmakers are eager to advance online safety legislation to address concerns about teen use of various social media and AI services, an AI preemption deal is not worth a package of unconstitutional, privacy-invading bills. The major bills involved in the grand compromise are rumored to include the App Store Accountability Act, the Kids Online Safety Act (KOSA), and the NO FAKES Act.
Meanwhile, the details of the AI framework portion of the deal are unknown at this time, but it sounds unlikely that the deal will comprehensively preempt state AI regulations. This could leave us with the worst of all worlds: “Swiss cheese” AI preemption that allows all existing state regulations to stand and many other mandates to follow while adding a new set of regulatory floors at the federal level that will just double down on the unconstitutional regulation of speech.
R Street has commented on these regulatory measures in recent work. Here are some of the problems with each measure:
KOSA
KOSA would likely be the most damaging of the three proposed bills. In the version passed by the Senate in 2024, KOSA would both put the government in charge of how social media platforms design their products and present content, and would indirectly force those platforms to enact age-verification measures. It also imposes a duty of care upon online platforms that would force them to take down any content that could conceivably promote harms as vague as “anxiety,” “depression,” or “encourage addiction-like behaviors by minors.” These provisions would place the government–led by the Federal Trade Commission–in charge of defining what kinds of speech violate this duty of care, and in doing so violate the First Amendment rights of both the platforms and their users.
App Store Accountability Act
The App Store Accountability Act would force app stores to verify the age of every mobile device user. For users under the age of 18, it would require that a parent or guardian be notified and give their consent for any new app download or major update. As we have repeatedly warned, both age-verification and parental-notification mandates have been found to be unconstitutional infringements on access to speech, and state-passed versions of this proposal are currently being challenged on First Amendment grounds.
NO FAKES Act
The NO FAKES Act is a bill that intends to protect against the fraudulent use of a person’s likeness copied by AI. However, it does so by creating a new intellectual property right that defines unauthorized use of a “digital replica” far too broadly and creates a notice-and-takedown regime for online platforms that host third-party content that provides little recourse to the person whose content is being taken down. This framework imposes broad liability that will almost certainly be abused to force sites to take down content such as satire, historical representations, political cartoons, and other protected speech.
Empowerment, Not Government Control
Of course, online safety is an important goal, but the manner in which we create that safety and the potential downstream effects of any actions must be considered seriously. While many in Congress want immediate regulatory action, it is important to realize that various tech companies have released many useful tools to address online safety concerns, and each month seems to bring expanded options. For example, Apple, Google, Meta, Microsoft, OpenAI, X.com, Snapchat, Discord, and others have recently rolled out new safety controls. These tools offer users and families a wide variety of ways to take control of their online experiences and AI use to determine what works best for their families. These approaches are preferable solutions compared to government-mandated regulations that undermine online freedom of speech. The measures under consideration also raise threats to user privacy by requiring that government officials take a more active hand in surveilling digital applications and platforms.
Bills like the App Store Accountability Act, KOSA, and the NO FAKES Act would also undermine online competition and innovation more broadly. By raising compliance costs and litigation risks, the measures threaten to undermine the vibrancy of digital markets.
If the White House and Congress swing for the fences with unconstitutional bills, they will end up in protracted litigation as the constitutionality of each measure gets challenged in court.
While some degree of state AI preemption is still needed, it should not be packaged as a compromise that undermines the First Amendment, parental rights, privacy, and online freedom more generally.