EFF on the California Governor's AI Executive Order: Start with the Real Risks

The Electronic Frontier Foundation (EFF) has responded to California Governor Gavin Newsom's executive order on artificial intelligence. The organization sees the document as an occasion for the necessary and thoughtful conversation about the technology and its potential harms, and calls it a «good start». In EFF's view, the state's residents feel a very real anxiety about AI, and California can lead the public discussion that is now overdue.

The central point of the statement, however, is that the most pressing threats have nothing to do with fantastical scenarios about a superintelligence slipping out of control. The problems are happening right now: biased algorithmic decisions in hiring and in the distribution of government benefits, AI-based surveillance systems (Flock cameras are cited as an example), and artificially inflated personalized pricing. These, EFF argues, are exactly the issues voters expect the authorities to address.

Reporting and «kill switches»

The organization supports the intention to expand the reporting requirements under SB 53 (2025) with respect to loss-of-control incidents, as well as the idea of independent third-party investigations. At the same time, EFF urges careful thought about how to make such investigations accessible to smaller developers too, not just the large labs.

A separate word of caution concerns the so-called «kill switches» for advanced AI systems. According to EFF, their effectiveness remains the subject of active research, so any regulation touching on cybersecurity practices in AI labs must be careful, precise and practical. What is more, emergency shut-off mechanisms controlled by the state risk becoming a tool of pressure and retaliation against legally protected speech. In support of this, the organization points to the Trump administration's actions against the company Anthropic earlier this year.

The priority is concrete harm

EFF's conclusion is blunt: genuine safety requires focusing on concrete, immediate and pressing threats, not on hypothetical catastrophes. The key condition is that algorithmic decision-making, in both the public and the private sector, respect people's rights and well-being. The organization calls on Newsom and state authorities to craft policy together with those who stand to suffer most from such technologies.

Источник: EFF Deeplinks