Meta Platforms on November 4, 2026 asked a federal court in California to stop Attorney General Rob Bonta from enforcing Assembly Bill 1709 against Facebook, Instagram and Threads, arguing that the law's ban on personalized feeds and autoplay for users under 16 violates the First Amendment and is preempted by Section 230 of the Communications Decency Act.

The complaint, filed in the US District Court for the Northern District of California as Case No. 3:26-cv-13201, runs to 49 pages and 157 numbered paragraphs. It brings five counts. Two rest on the First Amendment, and three on federal preemption. The law takes effect on January 1, 2027, eight weeks after the filing. The complaint is Meta's account of the law and of its own products. Every characterization of AB 1709 below is Meta's unless stated otherwise, and the State has not yet responded.

In Short

Meta has gone to court to stop a new California law that says children under 16 cannot be shown feeds that an algorithm picks for them, or videos that start playing on their own. Meta says choosing and arranging posts is a form of free speech, like a newspaper choosing which stories to print, and that a 1996 federal law stops states from punishing it for how it displays other people's posts. If the law stands, platforms that break it from January 1, 2027 could face fines of up to $50,000 for each affected child. Two things complicate Meta's case: a federal judge rejected a similar free-speech argument against an older California feed law in August, and many of the teen limits Meta now offers as better alternatives were set out in a settlement California agreed with Meta that same month.

What Meta filed

Meta Platforms, Inc. is the only plaintiff. The only defendant is Rob Bonta, "in his official capacity as Attorney General of California." The complaint is signed by Isaac D. Chaput of Covington & Burling LLP. Seven more Covington lawyers are listed, six of them with pro hac vice admission "forthcoming." It asks for the case to go to the San Francisco or Oakland division because Meta is headquartered in San Mateo County, where, according to the filing, "the impact of the Act on Meta, should it be deemed valid and enforceable, would be experienced."

The challenge is narrow by design. Meta brings what it calls an "as-applied challenge" covering only Facebook, Instagram and Threads, rather than asking the court to strike the statute down for everyone. It targets four provisions of the California Business and Professions Code: the definitions of covered features in section 22682(a)(1)-(2), the coverage definition in section 22682(c), the feature ban in section 22683, and the account-deletion rule in section 22684(b).

The five counts are:

  • Count One - personalized-feed ban, First Amendment
  • Count Two - autoplay ban, First Amendment
  • Count Three - personalized-feed ban, preempted by Section 230
  • Count Four - autoplay ban, preempted by Section 230
  • Count Five - account-deletion requirement, preempted by Section 230

The prayer for relief asks for declarations that the provisions are unlawful as applied to the three services, a preliminary and then permanent injunction stopping Bonta and his staff from enforcing them against Meta, and attorneys' fees under 42 U.S.C. section 1988(b). The document asks for a preliminary injunction only as part of the prayer for relief. It is not a separate motion, and no hearing date appears in the filing.

What the complaint leaves out of its targets matters as much. Section 22684(a), which according to the filing bars any covered service from providing covered features "to any user of any age unless the service implements an age verification system," is not among the challenged provisions. Nor is the penalty section, 22686(b).

What AB 1709 requires, according to Meta

The California Legislature passed AB 1709 on August 31, 2026, and Governor Gavin Newsom signed it on September 10, 2026. PPC Land reported the signing at the time. The complaint quotes Newsom's announcement that "[t]oday's legislation makes clear that California will not stand by while unregulated technology puts our children at risk," and notes that he called the package the nation's "strongest" rules on social media and online technology for minors.

The law regulates "covered platforms." A service is covered if it offers a covered feature "as a significant part of the service provided." As signed, the law names two such features. The first is a feed that the statute calls an "addictive feed," a phrase the complaint calls "pejorative," preferring the term "personalized feed." The second is autoplay. A third category is left open: "[a]ny other feature defined in regulations adopted pursuant to Section 22685," which lets the Attorney General add features by regulation. According to a footnote in the complaint, no such regulations had been adopted at the date of filing, and Meta "reserves its right to challenge any such regulations."

The statute defines the regulated feed as a service, or part of one, "in which multiple pieces of media generated or shared by users are, either concurrently or sequentially, recommended, selected, or prioritized for display to a user based ... on information provided by the user or otherwise associated with the user or the user's device." That phrase, "generated or shared by users," carries much of the case. A feed made of UGC is covered. A feed of material produced by the service itself is not.

Covered platforms may not provide any covered feature to a user under 16, and must "implement reasonable measures to ensure that [minors] are not provided any" covered feature. Two categories of service are excluded altogether: those where "interactions between users are limited to commercial transactions or to consumer reviews," and those that operate feeds "for the primary purpose of cloud storage."

The seven exemptions, as Meta reads them

The feed definition carries seven exemptions. The complaint walks through each with an example of how Meta thinks it would work in practice:

  • Information "not persistently associated with the user or user's device" that does not concern prior interactions. According to Meta, curation by a device's location might be exempt, while curation by a minor's stated home town might not.
  • Search terms "not persistently associated" with the user. A search for "San Francisco Giants" content might be exempt for a brief period, according to the complaint, but showing that content again outside the search might not be.
  • Privacy or accessibility settings, technical device information, or signals about whether a user is a minor. Meta reads this as allowing curation based on whether someone is a minor in general, but not on a narrower age band such as under 16.
  • Express requests for "specific media" or content from a specific user, provided audio or video is not played automatically. Following an account may be exempt; asking for a feed of "baseball" content may not.
  • Direct messages.
  • "A preexisting sequence from the same author, creator, poster, or source," again without autoplay. A single creator's posts in chronological order may be exempt. A chronological feed mixing two or more creators may not.
  • Whatever is "necess[ary]" to comply with the Act.

Deletion, age verification and penalties

Three further provisions round out the scheme. Under section 22684(b), a covered service must "delete the account of a user under 16 years of age and any personal information associated with" it, unless the user can keep an account that includes no covered features. Under section 22684(a), no covered features may be provided to anyone, of any age, without an age verification system. Where available, services must rely on age-bracket signals sent by app stores under California's Digital Age Assurance Act, Civil Code section 1798.500 and following, which also takes effect on January 1, 2027.

Civil penalties reach $50,000 "per affected minor" for knowing violations and $25,000 for negligent ones. Only the Attorney General or a local public prosecutor can bring an action.

Why does Meta think enforcement is coming? It points to the Assembly Appropriations Committee's May 12, 2026 analysis, which projected "[s]ignificant ongoing General Fund costs to the Department of Justice (DOJ) to ... bring civil enforcement actions against covered platforms." According to the complaint, the Legislature "anticipated not merely compliance with the Act, but active enforcement by the State against covered platforms." It adds that Bonta "has publicly pursued enforcement against Meta under other laws." Meta frames its standing on that basis: it says it has "an actual and well-founded fear" of enforcement, and that "Meta neither desires to nor intends to change its services to ban these practices for minors without regard to their (and their parents') individual choices and preferences, as the Act would require."

The First Amendment case

Meta's opening paragraph sets the register. AB 1709, it says, is "California's latest attempt to flout that fundamental principle and appoint the State as the editor, curator, and controller of online speech."

Feeds as editorial judgment

The core of Count One is the Supreme Court's 2024 decision in Moody v. NetChoice, which the complaint cites more than any other authority. Meta argues that its feeds "receive First Amendment protection" because they are the product of "expressive choices" about which user content to display and "how the display will be ordered and organized." In Meta's account, a recommender system is the machinery through which Meta's employees apply their editorial values at scale.

To make that case, the complaint describes in unusual detail how the feeds are put together. Curation, it says, "begins long before a user logs into their account." Meta's Community Standards are organized into 25 separate policies, from "coordinating harm and promoting crime" to "locally illegal content, products, or services." Content moderation, according to the complaint, "is not a separate step in creating a feed, but instead is infused throughout the process of generating a user's feed." Meta says it deprioritizes posts from frequent rule-breakers, clickbait and engagement bait, content it predicts users will report, and material similar to what users have said they do not want.

The complaint also states that Meta's ranking systems "are not based solely on predictions of what individual content item will be relevant to a user." According to Meta, they are also designed to push new content from small creators to progressively larger audiences, to mix posts from connections with recommendations, to avoid serving several videos in a row on the same topic or from the same creator, and to show less content users are likely to report.

The filing cites enforcement figures for the second quarter of 2026, when Meta says it took action on:

  • 6.9 million pieces of content on Facebook and 12.8 million on Instagram under its suicide, self-injury and eating disorder policy
  • 12.3 million on Facebook and 5.1 million on Instagram under its child sexual exploitation policy
  • 84 million on Facebook and 36.3 million on Instagram under its spam policy

Concrete examples of what would change

Meta offers examples of how it believes the ban would work. If a teen has shown interest in science experiments or local scholarships, the law "would preclude Meta from recommending additional content about science experiments or local scholarships to that user unless they specifically enter keyword searches for 'science experiment' or 'local scholarships' every single time they log in." The complaint lists California trade schools, universities, military recruitment, local volunteering, the San Francisco Chronicle and California public officials as further examples.

The complaint argues the law could also cut the other way. Meta says it might be barred from using a teen's history to avoid showing the same breaking news clip twice, and from deprioritizing "lawful posts concerning political or other controversial or sensitive topics after the user has indicated that she does not want such content recommended." It also argues that the law would stop Meta from honoring a teen's choice between the Limited Content and Ages 13+ settings when building a feed.

Dissemination and the autoplay "affirmative obligation"

Separately from editorial judgment, Meta argues the ban burdens its right to disseminate third-party speech, citing the Ninth Circuit's 1984 Cinevision decision on booksellers, theater owners and concert promoters as "a type of 'clearinghouse' for expression."

Count Two applies the same logic to autoplay. The complaint says the ban forces Meta "to make users take affirmative action before Meta can display expression," and likens that to permit requirements struck down in Watchtower Bible (2002) and to the 1965 Lamont ruling against a law requiring recipients of "communist political propaganda" to ask in writing for delivery. "The State cannot explain why playing content automatically would be harmful if the content were, for example, science experiments or math tutorials," the complaint states.

A footnote adds a detail that cuts in a different direction. "While Meta automatically plays some video content on its services," it says, "that feature is not itself a significant part of the Facebook, Instagram, or Threads services." In other words, Meta says its services are covered because of their personalized feeds, not their autoplay, and it attacks the autoplay ban as content-based for exactly that reason.

Content-based, speaker-based, and the Yelp comparison

Why strict scrutiny? Meta gives three reasons. First, the law covers feeds of user-generated content but not feeds of provider-selected content, so "video and music streaming services may be excluded" while Facebook, Instagram and Threads are not. Second, the commerce and review exemption means "commercial review websites like Yelp need not comply with the Act at all." Third, the ban targets personalized recommendations "while leaving generically popular content recommendations untouched," which Meta calls suppression of "speech bearing a particular quality." The complaint relies on district court rulings from 2023 to 2026 against social media laws in Virginia, Georgia, Ohio, Utah and Arkansas. It acknowledges that the Georgia and Ohio rulings were later reversed on other grounds by the Eleventh and Sixth Circuits.

The legislative record

The complaint spends several pages on the evidence the Legislature relied on. According to Meta, the Assembly Committee on Privacy and Consumer Protection's April 14, 2026 analysis identified two "primary risk factors" from the US Surgeon General's 2023 advisory: "harmful content" and "excessive and problematic use." The Legislature also cited, among other research, a synthesis of 20 studies on body dissatisfaction and disordered eating, research on "fear of missing out," and studies suggesting two-thirds of adolescents encounter "hate-based" content online.

Meta's argument is that every one of these is about content, not design. "The studies it cited do not identify harms from content-neutral design features," the complaint states. It quotes the Surgeon General's report as saying "[a]t this time, we do not yet have enough evidence to determine if social media is sufficiently safe for children and adolescents," and says the report does not mention personalized recommendations as a cause of harm. According to the complaint, "nearly all of the cited research 'is based on correlation, not evidence of causation.'"

Tailoring and alternatives

Even under intermediate scrutiny, Meta argues, the bans fail. The complaint calls them overinclusive because they apply "regardless of the minor's or even their parents' preferences and regardless of whether the service already voluntarily restricts minors' time spent." It calls them underinclusive because minors can watch the same videos after one tap, or on exempt services. "If the same expression may be viewed indefinitely after one additional click or may autoplay on an exempt service, the Act's chosen distinction bears no adequate fit to its asserted objectives," it argues. On time online, the complaint says "the way to address that problem is by limiting time spent online - not by regulating only personalized content recommendations."

The complaint does concede some ground: "the protection of children is an important governmental interest - an interest that Meta shares."

The Section 230 counts

Counts Three to Five rely on 47 U.S.C. section 230(e)(3), which provides that "[n]o cause of action may be brought and no liability may be imposed under any State or local law that is inconsistent with this section."

For the feed ban, Meta leans on two Ninth Circuit rulings. It quotes Doe 1 v. Twitter (2025), which it says held that Section 230 preempts state laws that "attack the manner in which Facebook's algorithm sorts, arranges, and distributes third-party content," and Doe 1 v. Meta Platforms (2026), which it says barred claims based on "promotion of engagement-driven [third-party] content through [a] recommendation algorithm," including "a personalized algorithm."

For autoplay, it argues that automatically playing media is a method of "distribut[ing] third-party content," a publisher function. It cites a 2023 ruling in the federal social media addiction litigation in the same district, which it says found claims based on Meta's "auto-play" feature barred by Section 230.

For account deletion, Meta cites a 2026 Ninth Circuit decision, Shared Partnership v. Meta Platforms, which the complaint says treated deleting a user's account page as a "quintessential editorial decision protected under Section 230." It also cites a 2025 ruling in this district, NetChoice v. Bonta, which it says found a separate California statutory requirement to delete certain user accounts preempted. The deletion rule also runs through the First Amendment counts. Meta says it is the State's indirect route to the same feature bans, and quotes the Supreme Court's Rutan ruling: "What the First Amendment precludes the government from commanding directly, it also precludes the government from accomplishing indirectly."

Meta's teen controls, as described in the complaint

A large share of the filing describes tools Meta already offers or says it will soon offer. These pages carry the less-restrictive-alternatives argument.

On Facebook and Instagram, teens aged 13 to 17 are placed by default in an "Ages 13+" content setting designed so that teens see content "generally similar to what teens would see in an age-appropriate movie." That covers both users who list a teen birthday and users Meta "proactively identifies as suspected teens." On Instagram, teens can choose a stricter "Limited Content" setting. With parental permission, teens on both services can move to a "More Content" setting aligned with ratings for ages 18 and over. On Threads, teens default to "See less" potentially upsetting content. PPC Land covered the move to movie-rating standards in October 2025 and the global expansion of the 13+ setting in June 2026.

The complaint then lists protections Meta says it is "in the process of rolling out" for California teens:

  • "more robust age-verification processes"
  • a default two-hour daily time limit, cumulative across Facebook and Instagram
  • a default block on nighttime access from 12 a.m. to 6 a.m.
  • push notifications muted by default between 8 a.m. and 3 p.m. on school days, defined as Monday to Friday between August 15 and June 15
  • a default block on seeing like and reaction counts

"Meta expects these defaults to be in place in the coming months," the complaint states. Parents and guardians can change them together with their teens, but teens cannot change them alone. Meta also says parents "will soon be able to change their teen's default settings to require non-personalized feeds and stop video and animation content from automatically playing." It says separate optional tools for parents and minors to turn off personalized recommendations and autoplay will arrive "in the coming months."

What the complaint does not mention

Three developments that PPC Land has covered bear directly on Meta's arguments. The complaint mentions none of them.

PPC Land reported on August 27, 2026 that Chief US District Judge Yvonne Gonzalez Rogers had signed a consent judgment between Meta and state attorneys general in the federal youth social media litigation, effective August 27. California is a party, through People of the State of California v. Meta, and carries the largest payment line in the settlement's schedule. The terms PPC Land reported line up closely with the complaint's list: a cumulative two-hour daily default across Meta platforms, a night mode from midnight to 6 a.m., school-hours notification muting from 8 a.m. to 3 p.m. on weekdays, like counts hidden by default, autoplay among optional protective settings, and a non-personalized feed option due within four months, meaning late December 2026.

The complaint presents these protections as Meta's own measures and cites them as evidence that AB 1709 is unnecessary. It does not say where they come from. At one point it describes Meta's services as among those that "already voluntarily" restrict minors' time spent. In October, Meta booked a $9.89 billion charge tied to the attorneys general teen settlement in its third-quarter results. The defendant in this new suit is the attorney general whose office is a party to that settlement.

The August ruling against Meta on SB 976

The complaint describes California's earlier feed law, SB 976, signed on September 20, 2024. That law restricts personalized feeds for minors but allows them with parental consent, and covers users through age 17. The complaint notes that Meta's challenge to it "is currently pending before the Ninth Circuit," citing an appeal docketed on August 11, 2026 as No. 26-5120.

What it does not describe is the ruling being appealed. On August 5, 2026, Judge Edward J. Davila of the same district denied preliminary injunctions sought by Meta, TikTok and Google against SB 976's feed and default-setting provisions. According to PPC Land's report, he found that ranking content by predicted engagement was not an expressive judgment, while treating content moderation choices as expressive. That is close to the theory at the center of Count One here. The complaint's long account of how Community Standards shape ranking, and its statement that Meta's feeds are not "feeds whose algorithms respond solely to how users act online," read as an attempt to put Meta's feeds on the expressive side of that line. Meta had sued over SB 976 in November 2025 with a single as-applied First Amendment count. This time it adds Section 230.

State courts on design features and Section 230

On Section 230 and autoplay, the complaint cites federal authority from the Ninth Circuit and the Northern District of California. Two state courts reached different conclusions this year in a different kind of case. In April, Massachusetts' Supreme Judicial Court held that Section 230 did not bar the Commonwealth's claims over design features, including infinite scroll and autoplay. In August, a New Mexico court ruled that Section 230 did not apply to the State's design-based public nuisance claim and ordered $567 million in abatement. That court refused remedies aimed at ranking, infinite scroll and autoplay. Both were liability cases rather than challenges to a statute, and neither binds a federal court in California. The Massachusetts case went back to the trial court.

California's record in court

Meta frames AB 1709 as the third round of a pattern. One heading in its background section reads: "California has already enacted several social media laws that have been found unconstitutional at least in part."

The first was the California Age-Appropriate Design Code Act, signed on September 15, 2022. According to the complaint, the Ninth Circuit held in 2024 that its data protection impact assessment requirement unconstitutionally compels speech, and in 2026 that its ban on using minors' data in ways "materially detrimental" to their well-being was unconstitutionally vague. The second was SB 976. The complaint states that SB 976's provisions on nighttime notifications and like counts without parental consent "have been deemed unconstitutional and preliminarily enjoined," the first by the district court in 2024 and the second by the Ninth Circuit in 2025.

AB 1709 differs from SB 976 in two ways the complaint highlights. It lowers the age threshold from under 18 to under 16. And it removes the parental-consent route, so a parent who wants a 14-year-old to keep a personalized feed would have no way to authorize it. That is the basis for the complaint's argument, citing Brown v. Entertainment Merchants Association, that the law imposes "governmental authority" in place of parental authority.

Why the marketing community is watching

The complaint does not use the word "advertising" in its argument. Its only reference to ads is a footnote: the figure that users share "over a billion stories and reels combined" every day is sourced partly to a 2024 Meta developer blog post titled "Unlock the Power of Reels Ads." Still, the features at stake are the ones that decide where attention goes on Meta's services, and with it where impressions are served. When AB 1709 was signed, PPC Land's coverage described the restricted features as the mechanism that produces addressable attention, with autoplay and personalized feeds as the surfaces where impressions are generated.

Four points in the filing matter for advertisers, publishers and creators.

Reach to under-16s. If the law takes effect as written, California users under 16 would see feeds ordered by the exempt methods Meta describes: followed accounts, single-creator sequences, non-persistent signals and in-session searches. Recommendations from accounts a teen does not follow, when based on that teen's activity, would fit none of the exemptions Meta describes. Meta says its systems use recommendations to push "new content from small creators to progressively larger audiences." The complaint gives no figure for how many California users this would affect.

Age verification for everyone. Section 22684(a) applies to users "of any age," according to the complaint, and Meta has not challenged it. A covered service could not show a personalized feed to an adult in California without first running that person through an age verification system tied, where available, to app-store age signals. Age assurance at the account level would sit upstream of every audience a platform builds in the state.

Who is exempt. Meta's speaker-based argument is also a map of the competitive field. Services built on provider-selected content, such as music and video streaming, may fall outside the law. So may commerce and review services. Platforms built on user posts are covered. If the coverage definition survives, how attention is distributed among those categories could shift for the youngest users.

User-directed ranking. In October, Facebook added a feature letting people type the topics they want in Feed and Reels. Under the exemption Meta describes for "express[] and unambiguous[] request[s]," a request for a topic such as "baseball" may not qualify, while following a named account may. The complaint does not mention that tool. Its argument implies that even ranking a user steers would remain restricted for under-16s.

The case also runs parallel to Europe. In July, the European Commission's preliminary findings under the Digital Services Act found that Meta had failed to assess the risks of Instagram's and Facebook's design, including highly personalized recommendations, autoplay and infinite scroll. California has written two of those three features into statute. The EU findings are preliminary.

What happens next

The complaint is the first document in the case, and the State has not yet responded. Several dates now overlap:

  • Late December 2026 - the four-month deadline under the August consent judgment for Meta's non-personalized feed option, according to PPC Land's reporting
  • January 1, 2027 - AB 1709 and the Digital Age Assurance Act take effect
  • Late February 2027 - the consent judgment's compliance date, according to PPC Land's reporting

The Ninth Circuit has yet to rule on Meta's SB 976 appeal. Its answer on whether engagement-ranked feeds are expressive is likely to bear on this case too. The Attorney General's power to designate further covered features by regulation remains unused, and Meta has said it may challenge any rule that follows.

Whether a ruling arrives before January 1 depends on a motion the complaint previews but does not include. Can a company argue that its default two-hour limit makes a state law unnecessary, when the state's attorney general is a party to the settlement that set that limit? The filing does not take up the question.

Timeline

Summary

Who: Meta Platforms, Inc., represented by Covington & Burling LLP, against California Attorney General Rob Bonta in his official capacity. The law affects Facebook, Instagram and Threads users in California under 16, and the platforms, creators and advertisers that reach them.

What: A 49-page complaint with five counts, seeking a declaration that AB 1709's bans on personalized feeds and autoplay for users under 16, and its account-deletion requirement, violate the First Amendment and are preempted by Section 230 as applied to Meta's three services, plus a preliminary and permanent injunction against enforcement. The law carries penalties of up to $50,000 per affected minor for knowing violations. Meta does not challenge the law's age verification requirement for users of all ages.

When: Filed on November 4, 2026. AB 1709 was signed on September 10, 2026 and takes effect on January 1, 2027.

Where: The US District Court for the Northern District of California, Case No. 3:26-cv-13201, with Meta asking for assignment to the San Francisco or Oakland division. The law applies to users residing in California.

Why: Meta argues that curating and displaying feeds is protected editorial activity under Moody v. NetChoice, that the law singles out user-generated content while exempting streaming, commerce and review services, and that Ninth Circuit precedent treats recommendation algorithms and account removal as publishing conduct protected by Section 230. The complaint does not mention the August 2026 consent judgment California signed with Meta, which sets out many of the teen defaults Meta cites as less restrictive alternatives, or the August 5 ruling that rejected a similar feed argument against SB 976.