A Mirror, Not a Message
On September 10, 2026, Google, Meta and TikTok asked the Ninth Circuit to block enforcement of California's SB 976 — the Protecting Our Kids from Social Media Addiction Act — while their constitutional appeal proceeds (MediaPost). The motion follows an August 5, 2026 ruling by U.S. District Judge Edward Davila, who denied the companies' preliminary-injunction bid and rejected their First Amendment theory outright. A personalized feed, Davila wrote, is "not an 'expressive' message; it is merely a mirror that reflects back to users their own perceived interests" (TheNextWeb) — engagement-driven ranking, in his view, is predictive modeling, not editorial judgment.
That distinction is the entire case. SB 976 bars platforms from serving minors an "addictive feed" — a stream ranked using the user's own behavioral data — without verifiable parental consent, defaults personalization off, and limits a minor's feed access to one hour a day by default unless a parent changes it (California DOJ). The Attorney General must finalize age-verification and consent regulations by January 1, 2027, and rulemaking is already underway. Meta's brief counters that the law "burdens Meta's First Amendment rights twice over," first by displacing its editorial judgment about how to present user speech (MediaPost). The companies lean heavily on Moody v. NetChoice (2024), where the Supreme Court held that platforms' curation, ranking and promotion of third-party content — even when automated — can itself be expressive activity protected by the First Amendment.
The Case for the Law
California's position deserves a fair hearing before it gets a rebuttal. Legislators built SB 976 around a real and well-documented phenomenon: engagement-optimized ranking systems are deliberately tuned to maximize time-on-app, and minors are the population least equipped to self-regulate against that design. The law doesn't ban any content — teenagers can still see everything they could see before, in chronological order. It targets the delivery mechanism, not the message, which is a meaningfully narrower intervention than the content-based restrictions courts have struck down in the past. And California isn't acting in a vacuum it created out of preference; it's filling one Congress left. The House passed its own youth-safety bill, the KIDS Act, 267–117 in June 2026, but stripped out the Senate's algorithmic-feed provisions entirely, leaving federal law silent on exactly the mechanism SB 976 addresses (Axios, June 29, 2026). Absent federal action, states are the only actors moving.
Why the Mirror Framing Doesn't Hold
Even so, Davila's "mirror" metaphor understates what a ranking system actually does. It doesn't reflect a user's interests; it selects, weights and sequences a subset of all available third-party speech according to criteria the platform designed. That is precisely the activity Moody protected — a point that will matter far more to three Ninth Circuit judges than to a district court applying rational-basis instincts to a genuinely sympathetic child-safety statute. If the panel treats ranking as expressive conduct, SB 976's default-off, one-hour, verified-consent regime looks less like a content-neutral manner restriction and more like a mandate to switch off a specific form of protected curation for an entire age class, regardless of whether any individual feed is actually harmful.
The law's bluntness compounds the problem. A one-hour default cap and mandatory default-off apply uniformly whether the content is contested political speech, a public-health explainer, or genuinely addictive slot-machine engagement bait — SB 976 makes no distinction. And the verification mechanism the law leans on to distinguish minors from adults carries its own costs: platforms must build systems to determine or verify age for users who never asked to be verified, creating friction and data-collection incentives that reach well past the population lawmakers meant to protect. The Electronic Frontier Foundation, in a September 17 amicus brief joined by the Center for Democracy & Technology and Wikimedia, made exactly this point, arguing SB 976 burdens teens' own speech and access rights and proposing a narrower alternative: require platforms to minimize the behavioral data they collect on minors, rather than banning the resulting curation outright (EFF).
That alternative points toward the proportionate outcome. Federal regulators have already shown a lighter-touch path is available: the FTC's February 2026 COPPA policy statement, adopted unanimously, encourages age-verification technology as a compliance tool rather than mandating architectural bans on how verified users receive content (FTC). A regime built on data minimization and default protections — rather than a categorical, one-size-fits-all feed ban enforced through a civil-penalty statute — would address the same engagement-design harms without asking judges to first decide that algorithmic curation isn't speech at all. The Ninth Circuit doesn't need to choose between protecting teenagers and protecting the First Amendment; it needs California to write a law narrow enough that it doesn't have to.