California Signed 13 AI Child-Safety Laws and a Body to Keep Updating Them

On September 10, 2026, California Governor Gavin Newsom signed 13 bills. The state calls them the strongest child-safety protections for AI chatbots and social media in the nation. The package reaches companion chatbots, physical toys with chatbots inside them, age verification for apps, and how K-12 schools handle student data and school-issued devices. One bill, AB 1709, also creates a new standing body: the e-Safety Advisory Commission.

Most of these bills set rules. AB 1709 builds something that keeps setting rules after this year's products are gone. That difference is the story.

The package reaches product design and legal liability

SB 1119 comes from Senator Steve Padilla, with Assemblymembers Buffy Wicks and Rebecca Bauer-Kahan. It requires safety protocols for AI companion chatbots used by minors. Time limits are part of it. So is surfacing mental-health resources inside the product, and telling a parent when a child turns off a safety setting. Independent child-safety audits and annual risk assessments come with legal liability attached.

SB 867, also from Senator Padilla, extends those rules to physical toys built around companion chatbots. AB 2, from Assemblymember Josh Lowenthal, adds civil penalties for injuries to children from covered platforms and products.

Assemblymember Wicks carried several more bills. AB 1856 and AB 2246 cover age verification for apps and children's access to online services and features. AB 1946 builds a reporting path for child sexual abuse material, and it expands that definition to cover AI-generated or digitally altered images.

Several of the 13 bills govern K-12 schools directly: student data, digital-wellness instruction, school-issued devices, and addictive feeds in school communications. These are K-12 measures. Higher education is not their subject.

A standing commission solves a problem no single statute can

Here is the design question underneath all of this. A statute fixes its own words on the day it is signed. Companion chatbot products, AI toys, and app defaults change every few months. Updating a statute takes a full legislative session.

That mismatch between the two speeds is what I mean by rules written for a slower world. Refreshing a law is slow by design, for good reasons, and that stays true even when the legislators writing it understand the technology well. They cannot cheaply reopen their own language every time a product category shifts under them.

Picture the simpler version of this package: one flat ban on companion chatbots for minors. It reads clean on signing day. It reads outdated within a year, the moment the same kind of model shows up inside a plush toy or a homework app. The authors of these bills saw that coming. AB 1709 answers with a body that has ongoing authority to keep looking, built into the same bill as its age-restriction rules. California built something designed to keep deciding. That is harder to build than a single rule, and it is worth far more once built.

Governor Newsom put the idea plainly when he signed the package:

Our children's safety deserves to be at the center of every conversation about technology. As innovation moves faster our protections must keep pace.

Keeping pace is the actual job here. A standing commission is one of the few tools built to do it.

The same instinct showed up two days earlier. On September 9, 2026, the same governor signed SB 813 and AB 1405. Those laws created a state AI-auditor registry and a verification framework, aimed at enterprise and institutional AI systems rather than children's products. Different subject, same move: build standing capacity instead of a fixed rule. I wrote about that one in California Just Built a Market for AI Auditors. The federal debate over AI oversight, which I covered in this piece on Congress and AI shutdown authority, keeps circling the same design question from a different angle.

The practical version of this arrives at a kitchen table

A parent hands a phone to a middle-schooler and hopes the defaults are sane. That is the whole policy debate, compressed into one ordinary moment. Most parents will never read AB 1709 or SB 1119. What they notice is smaller: does the session end on time, and does anyone tell them when a safety setting gets switched off.

First Partner Jennifer Siebel Newsom described the goal as safety "baked into design," not bolted on after. For anyone shipping software, that is a product requirement with a review date attached.

Teams that build for young users now have a cadence to run

Say your product, campus service, or agency tool reaches anyone under 18. The real change is the cycle itself. SB 1119 requires independent child-safety audits and annual risk assessments, with liability tied to the result. That cycle rewards the teams who start before anyone asks them to. It also gives the people already pushing for child-safety resources a concrete reason someone will fund the work.

Two questions are worth asking this quarter. When the e-Safety Advisory Commission issues guidance, who on your team reads it first? Who turns that guidance into a product requirement before the next audit? Answer those now, and a compliance exercise becomes a design advantage. The teams that can change a product fast will spend less on every future round of guidance than the teams that cannot. I write more about this in my work on AI governance.

California has committed to keep deciding on this. Everyone building for young users should take that seriously. The real work now belongs to product teams and to the institutions serving those same users, on both sides of one problem. When the commission's first guidance lands, how fast could your product actually change? I'm easy to find, and I want to hear your answer.

Here is what makes Alex a credible voice on this topic: Alex advises the California State University system on AI governance, which is where policy language like AB 1709 stops being abstract and becomes a set of decisions someone has to make about real products and real users.

Working through what California's child-safety rules mean for your product or institution? Book a conversation →