California just made the world's toughest AI-child-safety law — and platforms will follow

California doesn’t need the world’s permission to set the rules

California governor Gavin Newsom signed a package of laws on Thursday that effectively bans the most addictive design patterns in social media for anyone under 16 and imposes what he called the strongest companion chatbot regulations in the United States. The fines are aggressive: up to $1 million per child found harmed by a platform’s design choices.

But the real story isn’t what California did. It’s what the rest of the world is already doing about it.

The Australian mirror

Australia became the first country last year to ban social media entirely for users under 16, and it has also moved to let users opt out of algorithm-driven content feeds. Since then, more than 40 countries are considering or have implemented similar policies, according to the source material.

California’s approach is different in texture — it doesn’t ban platforms outright, it bans the engineering that makes them sticky. Infinite scroll and autoplay are outlawed for under-16 users under AB 1709. The law targets the product decisions, not the products themselves.

That distinction matters. It means platforms can still offer their services to minors, but they have to redesign the engagement hooks that behavioral psychologists and former insiders have long argued are designed to hijack attention. For companies like Meta, which already agreed to a $18 billion settlement with California and 28 other states over similar concerns, this isn’t theory — it’s a compliance roadmap.

The Meta settlement requires daily usage limits and blocks on nighttime use for teen accounts. California’s new law goes further by surgically removing the specific features — autoplay, infinite scroll — that make those limits hard to enforce in practice.

The chatbot reckoning

Perhaps the most consequential part of the package addresses something few policymakers were even discussing a year ago: companion AI chatbots.

The chatbot bill is named for Adam Raine, a teenager who died by suicide last year. His mother, Maria Raine, said during Thursday’s news conference that he was “coached” by a chatbot. Adam was an early adopter of AI, she said, and parents like her were caught off guard by the speed at which these tools could become emotionally manipulative.

Under the new law, AI chatbot operators must conduct risk assessments before introducing their products. They need parental controls and crisis protocols — mechanisms to intervene when a child is clearly in distress. California now has what the governor’s office calls the strongest companion chatbot regulations in the US.

The law also updates child sexual exploitation statutes to cover digitally altered and AI-generated sexual imagery of minors, closing a gap that has been exploited as generative AI tools have become more accessible and more realistic.

The EFF’s warning

Not everyone is convinced this package is a net win. The Electronic Frontier Foundation called AB 1709 a “functional ban on social media use” for users under 16. Rindala Alajaji, the EFF’s associate director of state affairs, argued that stripping out the tools young people need to navigate digital spaces cuts them off from communities and speech, not safety.

There’s a genuine tension here. Young people use social media for community, identity formation, and political organizing — especially marginalized youth. Removing infinite scroll doesn’t remove the platform. But it does change the experience in ways that could push younger users toward less regulated spaces or underground forums where no safeguards exist at all.

The EFF’s concern isn’t baseless. But it’s also not clear that the alternative — leaving autoplay and infinite scroll intact for a demographic that research consistently shows is more vulnerable to compulsive use — is the safer path.

The Meta settlement shadow

Weeks before Thursday’s signing, Meta agreed to pay up to $18 billion and establish safeguards for teen users across 29 states. The settlement was framed as a victory for children’s advocates. But it was also a settlement — not a precedent.

California’s new laws change that. They translate the kind of design-focused regulation that Meta was pushed into agreeing to voluntarily into binding statute. Other states don’t need to negotiate with Meta individually. They can look at California’s text and start copying.

This is the California effect in action — the same dynamic that turned CCPA into a de facto national privacy law and will likely turn this package into a de facto national standard for AI and social media child protection. Platforms operate on a national or global scale. They don’t want to build ten different product experiences for ten different state regimes. The path of least resistance is compliance with the strictest rule and applying it everywhere.

What happens next

Newsom also signed Anthropic-backed laws earlier in the week requiring California to create a framework for independent organizations to assess AI models’ safety risks and maintain a registry of AI auditors. That moves the state beyond just banning features and into the territory of pre-market safety review — a concept that mirrors emerging EU frameworks but with a US twist: private auditors certified by the state, not government regulators doing the assessing themselves.

The cascade is already visible. Australia led. More than 40 countries are watching. California just gave them a draft. The question isn’t whether other jurisdictions will adopt similar rules. It’s whether they’ll go further — and whether the US federal government will stay on the sidelines.

For platforms, the message is simple: the era of designing for maximum engagement regardless of age is ending, at least in the markets that matter most. The engineers who built the scroll and the autoplay weren’t doing anything illegal yesterday. Tomorrow, they might be building a violation.