California passes sweeping AI and social media restrictions as tech's home state tightens grip on platforms and automation

The shape of the package

California legislators passed 26 bills covering AI and social media on Monday 1 September 2026, the final day of the legislative session. Governor Gavin Newsom has until 30 September to sign or veto them, and his choices will shape not only how the world's fifth-largest economy regulates tech, but how the rest of the US follows.

The bills break into three broad groups: restrictions on how platforms serve young people, protections for workers facing AI-driven displacement or algorithmic management, and attempts to control the infrastructure costs and energy impact of AI data centers. Together, they represent the most ambitious state-level attempt to regulate platform design and AI deployment in the US to date.

What makes the package politically striking is its timing. The social media bill passed just a week after Meta agreed to pay $18 billion to settle claims that its products harm children, and the worker protection measures arrive as AI-driven layoffs and hiring freezes become harder to ignore. California is legislating with fresh evidence in hand.

Social media: banning the features, not the apps

AB 1709 would prohibit social media platforms from providing "addictive features", including algorithmically recommended content and videos that autoplay, to users under 16. If signed, the law will also ban endless scrolling for the same age group.

The bill would amend California's existing Protecting Our Kids From Social Media Addiction Act (SB 976), a 2024 statute that requires parental consent before platforms can serve algorithmic feeds to under-18s. That earlier law is already being challenged in court by Google, Meta and TikTok on First Amendment grounds, and AB 1709 will almost certainly face the same legal obstacles.

The bill establishes an e-Safety Advisory Commission to advise on implementation, housed within the California Department of Justice. The commission will have the authority to adjust the definition of "covered platform" as new products emerge, giving the state a mechanism to expand the law's scope without returning to the legislature.

The distinction being drawn here is crucial for social marketers. The bill only restricts platforms that use addictive design features; young people will still be able to use social media applications that rely on non-addictive user interfaces. In practice, that means chronological feeds, user-initiated search, and following-based timelines remain accessible. What disappears for under-16s is the For You page, the Explore tab, and anything that decides what you see next without you asking for it.

That creates a two-tier design challenge for platforms: serve the same product to all users and lose the under-16 audience in California entirely, or build age-gated versions where younger users see fundamentally different interfaces. Either way, campaigns targeting or including younger demographics will need to rethink distribution assumptions. Reach through recommendation is about to become adult-only in the state with the largest economy in the US.

The bill's opponents argue it amounts to a functional ban. Critics say the law would prohibit platforms from offering virtually every functional recommendation algorithm to anyone under 16, which are in reality the basic tools services use to identify what user-generated content someone might want to see. Enforcement will hinge on age verification, and California has yet to clarify exactly what "reasonable measures" platforms must take to confirm a user's age without creating privacy risks for everyone else.

Worker protection: no firings by algorithm alone

SB 947, known as the No Robo Bosses Act, bars California employers from relying solely on automated decision systems (ADS) to fire or discipline workers. When ADS are used to assist in termination or disciplinary decisions, the bill requires human oversight and verification.

The bill also prohibits the use of ADS that rely on workers' personal information to predict their future behaviour. That provision targets the emerging use of "predictive analytics" in workforce management, tools that score employees on their likelihood to quit, unionise, or file complaints.

If an employer primarily relies on an ADS to make a disciplinary or termination decision, it must direct a human to corroborate the decision and provide a post-use notice to the affected worker. The bill does not specify how detailed that notice must be or whether it must disclose the ADS vendor, though that will likely be clarified in regulations if the law is signed.

For social media teams, especially those working in-house at large organisations or managing freelance and contractor rosters through automated systems, the practical question is whether performance tracking, scheduling or workload assignment tools fall under the definition of ADS. The statute defines them broadly as any system that uses computation to assist or replace human decision-making, which could capture platform-native analytics dashboards if they're used to justify headcount decisions.

The bill has strong backing from organised labour but faces resistance from business groups who argue it will slow down legitimate workforce management and create compliance burdens without meaningfully protecting workers. Governor Newsom vetoed SB 947's predecessor, SB 7, in 2025, citing concerns about scope and overlap with existing employment law. Whether the revised version survives is uncertain.

Technological displacement: 90 days' notice before AI takes the job

SB 951, the California Worker Technological Displacement Act, would require businesses to give at least 90 days' advance notice before AI-driven layoffs affecting 25 or more workers or 25% of the workforce, whichever is less.

That is 30 days longer than the standard 60-day requirement under California's existing WARN Act, and the trigger threshold is drastically lower. Traditional WARN applies to employers with 75 or more employees laying off 50 or more people; SB 951 drops that floor to 25 workers or a quarter of the workforce.

The notice must contain detailed information including the specific AI system used, the vendor that developed or sold it, the job functions being automated, the justification for the technology's adoption, and whether retraining opportunities are available. The bill prohibits discharging affected workers during the 90-day period following the initial notice.

For agencies and in-house teams evaluating AI tools that could automate content production, community management, or campaign reporting, SB 951 creates a new layer of workforce planning complexity. Implementing a generative tool that displaces even a small team now triggers mandatory disclosure, a three-month waiting period, and potential penalties if the process is mishandled.

The law also introduces a "technology hiring disruption notice" requirement. Employers must provide written notice when they permanently stop hiring for an occupation or position because AI or automation has taken over that work, sent to the Employment Development Department and the local workforce investment board. That creates a public record of which roles are disappearing and why, data that could become politically sensitive if concentrated in specific industries or regions.

Data centers: making AI infrastructure pay its share

SB 1168 directs the California Public Utilities Commission to examine data centers' energy use and ensure that data centers cover their fair share of transmission and distribution costs to prevent their growing electricity demand from driving up rates for residents.

The bill won unanimous bipartisan support, a rarity in California's polarised legislature and a sign that energy cost anxiety cuts across party lines. California already hosts one of the nation's largest concentrations of data centers, with demand expected to grow by 2.3 gigawatts by 2030.

The bill stops short of mandating specific rates or infrastructure charges. Instead, it requires the commission to assess opportunities for rate structures that prevent cost-shifting onto households and small businesses. That gives regulators room to design solutions but also delays any concrete relief until those studies conclude and new tariffs are proposed, a process that typically takes years.

For the social media marketing industry, the relevance is indirect but real. The platforms marketers depend on run on data center infrastructure, and if California succeeds in shifting more of that cost onto the operators, it creates a model other states will watch closely. Higher infrastructure costs eventually surface in platform pricing, ad rates, or reduced investment in features that don't generate direct revenue.

The broader data center package passed on Monday also included bills requiring environmental impact reports, energy and water usage reporting, and contributions to grid maintenance. Taken together, they represent California's attempt to slow the infrastructure land grab happening as AI companies race to secure compute capacity, often with little transparency about environmental or fiscal impact.

What happens if Newsom signs them all

If the governor approves the full package, California becomes the first state to regulate platform design for minors, mandate human oversight of algorithmic employment decisions, require advance notice for AI-driven job losses, and force data centers to internalise their grid costs. That would establish a compliance floor for any company operating at scale in the US, because building separate products and processes for California alone is rarely viable.

The alternative is a wave of legal challenges. The 2024 social media law is already being challenged by Google, Meta and TikTok, who argue it violates their First Amendment right to editorial control over their platforms. AB 1709 will face identical arguments, and the worker protection bills will almost certainly be tested on grounds of federal preemption and vagueness.

Newsom has until 30 September to decide. His choices will reveal how far California is willing to go in regulating an industry it also depends on for tax revenue, and whether the state believes the political cost of inaction now outweighs the litigation risk of signing bills the courts may eventually strike down.

For social media professionals, the immediate priority is scenario planning. If AB 1709 becomes law, how does your content strategy adapt when algorithmic distribution to under-16s disappears? If SB 947 passes, which of your workforce tools might require human sign-off before they can be used to manage performance or headcount? And if SB 951 takes effect, what does your AI adoption roadmap look like when displacement triggers a 90-day public notice and a ban on layoffs during that window?

The answers depend on what Newsom signs. But the questions are now unavoidable.

How this connects to the EU's approach

California's move follows months after the EU began enforcing eight-hour data handover rules and weeks after the state proposed fines for creators who don't disclose paid political posts. The pattern is consistent: states and blocs are no longer waiting for federal frameworks. They are building their own, and the compliance burden is fragmenting by jurisdiction.

The difference is enforcement. The EU's Digital Services Act carries penalties of up to 6% of global revenue. California's bills rely on state-level civil actions and attorney general enforcement, which means smaller budgets, slower cases, and less certainty about how aggressively the laws will actually be applied. That creates risk for platforms, but also room to test the boundaries in ways that would be far more costly under Brussels' rules.

For marketers, the result is the same: you are now operating in a regulatory patchwork where the rules governing reach, targeting, data access and disclosure vary not just by country, but by state. The days of assuming one set of platform mechanics applied everywhere are over.