• by Wendy Davis , February 23, 2026

California's restrictions on online recommendations do not violate the First Amendment, state Attorney General Rob Bonta argues in court papers filed late last week.

He is urging U.S. District Court Judge Edward Davila in San Jose to reject requests by Google, Meta and TikTok to block the Protecting Our Kids From Social Media Addiction Act (SB976) -- a 2024 law that prohibits social platforms from algorithmically recommending posts to minors without their parents' consent.

As its name suggests, the law aims to prevent teens from becoming "addicted" to social media platforms that could serve them harmful content -- such as posts encouraging dangerous activity. Several key provisions of the law won't take effect until next year, at the earliest.

The platforms argued in papers filed last year that the law violates their First Amendment rights to make editorial decisions regarding content.

For instance, Google argued that YouTube's recommendations "reflect its own editorial judgments about the videos it believes will be relevant, valuable, and enjoyable for each particular user at a particular moment."

Meta added that the law would hinder its ability to curate speech.

"If a teen demonstrates an interest in science experiments or content about local scholarships, the Act would preclude Meta from recommending additional content about science experiments or local scholarships to that user unless they specifically search for 'science experiment' or 'local scholarships' every single time they log in," the company wrote.

Bonta counters that algorithmic recommendations are not the type of "speech" that's protected from governmental censorship.

"First Amendment protection extends only to human beings," he argues. "Where, as here, humans are so far removed from the computerized processes by which plaintiffs’ personalized feeds are compiled, and no human understands those processes or knows what content is actually being delivered, First Amendment protection does not apply."

Bonta says evidence that's come to light in court proceedings has shown that "humans do not write the rules that govern how feeds are personalized," and that "the algorithms write, and constantly rewrite, their own rules."

The algorithms "do not reflect the expressive choices of any actual human being -- in fact, the humans at plaintiffs’ companies cannot predict what the algorithms will 'say,' and the algorithms often promote user posts that their human creators desire not to promote," Bonta argues.

"It is of no moment that plaintiffs’ algorithms were originally created by humans," he adds. "The fact that a mechanism might be created by humans does not derivatively mean that any output created by that mechanism is the product of a human, much less expressive speech by a human."

The tech industry group NetChoice previously sued to block the law, but the 9th Circuit Court of Appeals ruled in September that NetChoice wasn't the right party to bring suit. The appellate judges said in the ruling that only individual social media companies could sue over the restrictions.

Davila is expected to hold a hearing in the matter on May 7.