EFF: California’s AI Executive Order is a Good Start, But Must Focus on Immediate Harms

California Governor Gavin Newsom has issued an executive order establishing a state-level dialogue on potential artificial intelligence harms, earning a cautious welcome from the Electronic Frontier Foundation. The organization supports expanding SB 53 reporting requirements for loss-of-control incidents. However, it emphasizes that immediate regulatory focus must center on tangible issues like algorithmic bias, pervasive surveillance, and personalized pricing.

California Governor’s Executive Order Draws Cautious EFF Welcome

Prioritizing Operational Realities Over Speculative Scenarios

Everyday Californians are experiencing genuine anxiety regarding the rapid deployment of artificial intelligence.

Rather than getting bogged down in speculative, science-fiction scenarios concerning rogue super-intelligence, technology policy must grapple with operational realities happening right now. Biased algorithmic decision-making actively impacts employment applications and government benefits distribution. Meanwhile, AI-powered surveillance systems, such as Flock cameras, alongside artificially inflated personalized pricing models, create immediate friction for civil liberties.

State and federal leaders face mounting public pressure to enact sensible guardrails. Today’s executive order serves as a constructive starting point for addressing these public anxieties, provided that subsequent policy implementation directly protects consumer rights and human well-being.

Expanding Reporting Mandates and Managing Kill Switch Risks

A central element of the state’s recent directive involves expanding the reporting requirements established under SB 53 (2025), specifically targeting loss-of-control incidents and mandating third-party investigations. The EFF has voiced strong backing for these investigative frameworks while simultaneously urging the Government Operations Agency to ensure these reporting structures remain accessible to smaller developers.

At the same time, technical pragmatism must guide cybersecurity mandates. The effectiveness of kill switches in advanced AI systems remains a complex area of active research rather than an out-of-the-box utility. Regulators targeting lab security practices must maintain precision and operational practicality.

Guarding Against Political Retaliation and Civil Liberties Threats

Government-controlled kill switches carry an inherent structural risk: they can easily be co-opted as tools for political retaliation against protected speech. This danger was clearly illustrated earlier this year when the Trump Administration took retaliatory actions against Anthropic.

Photo: europesays.com

True safety in the digital age requires California to collaborate directly with communities most vulnerable to algorithmic discrimination. By keeping the policy framework anchored to concrete, urgent harms rather than abstract threats, the state can build a human-centered regulatory model that balances technical innovation with fundamental civil rights.