EFF Statement on California Governor's Executive Order on AI

Governor Gavin Newsom issued an executive order aimed at curbing the most pressing harms of artificial intelligence in California, a move that forces policymakers, companies, and citizens to confront how AI already reshapes everyday life. The order is not about futuristic super‑intelligence; it targets concrete issues like biased hiring algorithms, AI‑driven surveillance, and price discrimination that affect ordinary residents now. Understanding the order’s scope helps anyone who relies on state services, shops online, or worries about being watched.

Current AI harms affecting Californians

AI systems are already influencing who gets a job or government benefits, with biased algorithmic decision‑making leading to systematic exclusion of marginalized groups. When an algorithm flags a worker as “high risk” or denies a housing subsidy, the lack of transparency makes it nearly impossible to contest the outcome.

Beyond employment, the state is seeing the rollout of Flock cameras, AI‑powered surveillance that can track individuals in public spaces and feed data to law‑enforcement databases. This technology amplifies existing privacy concerns by turning everyday movement into a searchable record, raising the specter of a surveillance state at the municipal level.

Policy levers in the executive order and pending legislation

The order expands reporting requirements under SB 53 (2025), mandating that AI developers disclose “loss‑of‑control” incidents and submit to independent audits. By tying these disclosures to third‑party investigations, the state hopes to create a factual record that can be used to hold developers accountable.

EFF urges the administration to make these investigations accessible to smaller developers, preventing a regulatory gap where only large firms can afford compliance. The order also calls for research into the feasibility of “kill switches,” though it cautions that any mandate must be narrowly tailored to avoid unintended technical fallout.

Risks of overbroad regulation and retaliation

While kill switches sound like a safety net, government‑controlled kill switches could be weaponized to silence dissent, as illustrated by the Trump Administration’s retaliatory move against Anthropic earlier this year. Such precedents show that technical controls can become tools of political repression if not carefully constrained.

Overly prescriptive cybersecurity rules for AI labs risk stifling innovation and may inadvertently create new vulnerabilities. Effective regulation must balance the need for safety with the practical realities of AI development, ensuring that safeguards do not become blunt instruments for censorship.

What This Actually Means For You

1. Algorithmic decisions affecting employment, benefits, or credit may be biased; you have a right to request explanations and contest outcomes under state law.
2. AI‑enhanced surveillance like Flock cameras can capture your movements in public; be aware of where such systems are deployed and consider privacy‑focused community actions.
3. Personalized pricing driven by AI can lead to higher costs for certain demographics; monitor price changes and report suspicious patterns to consumer‑protection agencies.
4. Reporting requirements under SB 53 will generate public data on AI incidents; use this information to advocate for stronger oversight in your local jurisdiction.

Immediate Action Steps

Participate in the public comment period for the Government Operations Agency’s recommendations; your feedback can shape how third‑party investigations are structured and who gets access to them. Submit concrete examples of AI‑related harms you’ve experienced to the state’s reporting portal to ensure they are captured in the new data set.

Stay informed about where AI surveillance is installed in your community and join local coalitions that demand transparency and oversight. By organizing collectively, residents can pressure municipalities to adopt clear signage and opt‑out mechanisms where feasible.

Frequently Asked Questions

What does California's executive order on AI require from developers?

The order expands the SB 53 reporting framework, obligating developers to disclose any loss‑of‑control incidents and submit to independent, third‑party investigations, with a focus on making these processes available to smaller firms.

How does AI‑powered surveillance like Flock cameras affect privacy?

Flock cameras use AI to analyze video in real time, creating detailed records of individuals’ movements that can be stored and cross‑referenced with other data sets, effectively eroding anonymity in public spaces.

Can government‑controlled kill switches be used against protected speech?

Yes; the order warns that kill switches could be misused as a tool of retaliation, citing the Trump Administration’s action against Anthropic as an example of how technical controls can suppress dissent.

What Do You Think?

Should California prioritize immediate, concrete AI harms over speculative future risks, or attempt to regulate both simultaneously?