California puts autoplay and algorithmic feeds behind an under-16 check

A feature ban is enforced per account, on the serving path. AB 1709 reaches autoplay and algorithmic feeds built around a user's history and profile ^[11], and the law as signed also covers infinite scroll and algorithmic recommendations ^[3]. In most codebases that is three separate switches: a ranking path that ignores history and profile, pagination where the infinite list used to be, and autoplay off for the account. Each has to be selectable at request time, so the under-16 answer has to be available wherever the feed is assembled.

The definition is where the difficulty sits. California defines addictive features as ones that are "psychologically exploitative" ^[3], which is a legal standard and not a list of engineering constraints. Compliance with a standard like that gets proved with documents. For companion chatbots the package says so directly, requiring risk assessments before a new or substantially modified product ships, plus independent compliance audits ^[14], with audit findings submitted to the California Attorney General ^[18]. Newsom's office describes the package as including the nation's first requirements for independent child safety audits and annual risk assessments for companion chatbots ^[13]. The auditors got their own statutes a day earlier: SB 813 sets a framework for independent organizations to assess AI systems' compliance with state law, and AB 1405 creates a registry of AI auditors with standards for their independence ^[23][24]. How "substantially modified" gets read sets the release cadence for a team that ships model updates weekly.

Newsom stressed the mechanics. "This is about the features themselves. This is about actually addressing the problem, the scrolling, the algorithms," he said, adding that "if you address that, then the age issue is not the essential issue" ^[5]. The chatbot law still needs to know the user's age. According to the bill's authors, operators must use age-bracket signals supplied through operating systems ^[15], and the same day's package included laws on age verification ^[28]. The age check moves into a dependency: the platform reads a bracket the operating system hands it.

Chatbot defaults are named in the statute too, including usage limits, disabled notifications and restrictions on persistent conversational memory, with changes reserved for parents ^[16]. Since the change goes through a parent, the permission model differs from one the user controls.

Meta had already agreed to a similar approach as part of a settlement with California and dozens of other states that sued over children's privacy and consumer protection claims ^[8]. It is arguing against the feed rules anyway, telling the Associated Press that personalization helps it give teenagers relevant, age-appropriate content ^[20]. "Personalization is also how we deliver age-appropriate content for teens that is relevant to them," spokesperson Jim Cullinan said ^[21]. Taken as an engineering claim, that puts age-appropriateness inside the ranking model; the ban forces it into classification and eligibility filters that run before ranking. Snap, TikTok and Google-owned YouTube did not reply to requests for comment ^[9].

For large platforms the penalties run up to $1 million per child for negligently causing harm ^[12]. A finding covering a thousand children comes to a billion dollars ^[25]. The reports do not say when AB 1709 takes effect ^[27]. California is betting that the per-child penalty forces compliance where Australia's ban has not: it took effect in December, and a July report from Australia's own government found enforcement spotty with many children still using social media ^[7].