• by Wendy Davis , 4 hours ago

California is urging a federal appellate court to reject tech companies' request to block a state law that prohibits social platforms from algorithmically recommending posts to minors under 18 without the consent of a parent or guardian.

"Peer-reviewed scientific research confirms that algorithmically generated personalized feeds are habit-forming, addictive, and cause an array of harms to children, including poor sleep, negative self-image, anxiety, and depression," California Attorney General Rob Bonta argues in papers filed Tuesday with the Ninth Circuit Court of Appeals.

"There is good news, however," Bonta adds. "Limiting exposure to these addictive mechanisms can reduce those harms and improve children’s well-being."

His argument comes in response to requests by Google, TikTok and Meta to block the restrictions in the Protecting Our Kids From Social Media Addiction Act (SB976). The tech companies say the law is unconstitutional, arguing their algorithmic recommendations are protected by the First Amendment because they reflect editorial decisions about how to curate speech.

The fight over the law dates to 2024, when the tech industry group NetChoice claimed in a lawsuit that the statute violates web publishers' First Amendment right to recommend content, and teens' constitutional right to access content.

Last year, the appellate court said NetChoice wasn't in a position to make that argument. Instead, the appellate court ruled, individual social platforms that want to block the law would have to sue individually.

Google, Meta and TikTok responded by alleging in separate complaints that their recommendations were editorial speech, protected by the First Amendment.

U.S. District Court Judge Edward Davila in the Northern District of California rejected the platforms' arguments in August, writing that personalized recommendations aren't the type of "expressive" speech protected by the First Amendment.

The platforms have appealed that ruling and are asking the Ninth Circuit to block enforcement while their appeal proceeds. They argued in papers filed last month that their algorithmic feeds reflect their editorial decisions.

For instance, Google argued that YouTube's personalized recommendations "reflect its expressive choices to provide tailored, age-appropriate content to minors in different age groups and to avoid repeatedly recommending to teens content that, while innocuous as a single video, could be problematic if viewed in high quantities."

"When YouTube’s algorithm determines that a sports video -- rather than a math video -- will be interesting to a user who has watched physics videos, that is because YouTube engineers decided that physics and math are not sufficiently diverse topics and trained the algorithm accordingly," YouTube contended.

Bonta argues in his new papers that the algorithms' purpose is to keep users on the apps for as long as possible.

"The only things plaintiffs’ feed-personalization algorithms can perceive, and try to predict, are numeric data metrics, like the amount of time a user spends on a platform and how much the user engages with content (e.g., watches videos to completion, comments on them, reshares them, etc.)," he argues.

"Plaintiffs’ algorithms track user-engagement metrics, like the amount of time a user spends on posts, as 'proxies' for assessing which posts are 'interesting,' 'relevant,' or 'valuable' [to] that user," he writes. "Simply put, 'interesting,' 'relevant,' and 'valuable' reduce to, whatever will get that user to spend more time on the platform."

Google, Meta and TikTok are expected to file additional arguments by Nov. 3.