EFF Issues Statement on California Governor's AI Order

As reported by Rindala Alajaji in EFF Deeplinks, the Electronic Frontier Foundation has published an official statement in response to the executive order on artificial intelligence by California Governor Gavin Newsom. It was published on September 18, 2026. While welcoming the state-led dialogue, the statement called for addressing real-world harms rather than fanciful superintelligence. Bias in employment and benefits decisions, surveillance technology, and pricing issues were at the center.

The executive order will be an opportunity for debate over the potential harms of AI. Strong anxiety about the dangers of AI is spreading among state residents. As an organization that aims to empower people through technology, EFF welcomed the state’s leadership. Calls for action by state and federal leaders are growing. The executive order was positioned as a starting point.

Overview and Aims of the California

Governor’s Order The executive order encourages thoughtful dialogue on the dangers of AI. It outlines a framework for the state to take the lead in addressing concerns. EFF received this direction as a necessary discussion. Behind it lies growing anxiety among state residents. AI has increasingly penetrated employment, public services, and consumer life. State authorities are rushing to build the foundation for policy. At the federal level, debate over AI regulation continues. California is home to a concentration of technology companies. The state’s decisions will influence other states and federal policy. The order asked relevant agencies to develop recommendations. The California Government Operations Agency will prepare proposals to the governor. The content of the proposals could be reflected in future legislation.

Immediate Concrete Harms Highlighted by EFF

EFF did not treat a science-fiction-style runaway superintelligence as the main concern. It said priority should be given to addressing harms that are already occurring. First, there are biased automated decisions in employment and public benefits. Unfairness could arise in hiring and benefits screening. Second, there is the expansion of surveillance systems using AI. As a specific example, the surveillance network of Flock cameras was cited. The spread of license plate readers is sparking debate in many places. Third, there are unjustified price increases through personalized pricing. Price manipulation based on individual attributes and behavioral history was highlighted as a problem. These are not abstract future forecasts. They are events already affecting people’s daily lives. EFF called for institutional design to protect rights and welfare. Automated decisions by both government and the private sector are covered. It also urged collaboration with the people most vulnerable to harm.

As an example of knowledge-sharing for developers, there are also initiatives such as Stack Overflow for Agents Beta Release: AI Agent Knowledge Sharing. A foundation is needed to increase the transparency of automated decisions. A system for verifying surveillance technology will also be a challenge. Disclosure of pricing mechanisms will also be an issue. The state needs to establish effective remedies.

Strengthening Reporting Obligations and

Third-Party Investigations The statement supported expanding reporting requirements under SB 53. SB 53 is a state law enacted in 2025. It set out a policy of broadening reporting of loss-of-control incidents. Loss of control refers to behavior beyond developers’ intent. It also supported combining this with third-party investigations. Verification by outside eyes improves safety. EFF called for consideration for small and medium-sized developers. It said mechanisms are needed to make third-party investigations accessible. A system that only large laboratories can handle would be insufficient. Costs and staffing burdens could fall disproportionately on smaller businesses. An open verification system supports trust across the industry. Standardization of reporting will also be important. Operations that use reports to prevent recurrence will be tested.

Kill-Switch Debate and Free Expression Concerns

In the context of the executive order, debate over shut-off devices emerged. The effectiveness of shut-off functions in advanced AI systems is still under research. EFF emphasized this point in its advice to the Government Operations Agency. Regulation of AI laboratories’ safety measures should be cautious. It should be limited to precise and practicable measures. Overly broad obligations could hinder development. Requirements designed to reflect the reality of the technology are needed. It also expressed caution about government-controlled shut-off devices. There is a danger they could be used for retaliation against protected speech. As an example, it cited retaliatory measures by the Trump administration against Anthropic earlier this year. Safety features could become a means of political pressure. Review is needed from the perspectives of separation of powers and freedom of expression. Technical shut-off feasibility and institutional prevention of abuse are separate issues. The two need to be designed separately.

Focus of State Safety Measures and Future

Policy Challenges EFF clarified the conditions for true safety. It said the focus should be on concrete and imminent harms. Mechanisms are needed to ensure automated decisions respect people’s rights. In the public sector, accountability and correction procedures are essential. In the private sector, audits and disclosure of bias are challenges. Surveillance technology requires limits on use and oversight. Pricing requires transparency and improved complaint handling. Governor Newsom and state authorities were asked to form policy with deliberation. Collaboration with vulnerable parties is a prerequisite. Rather than blanket bans, responses tailored to actual harm are effective. The state’s practices will become a reference for other regions. They could also affect legislative debate in Congress. The intersection of technology and legal systems is being questioned anew.

Editorial Opinion

On short-term impact: in the next 3 to 6 months, review of SB 53 implementation and preparation of reporting formats are expected to advance. Discussions over who will conduct third-party investigations and bear the costs are expected to intensify. Without support measures for small and medium-sized developers, disparities will emerge in the verification system.

On the long-term perspective: in 1 to 3 years, audits of employment screening and benefits decisions are expected to become institutionalized. Disclosure rules for surveillance cameras and price calculations are expected to be linked to consumer protection. Standardization of shut-off functions is expected to proceed gradually depending on research progress.

Questions from the editorial team: where is the boundary between government-controlled shut-off functions and freedom of expression. What is needed to balance prevention of actual harm with technological innovation. How to reflect the voices of vulnerable people in institutions is being questioned.

References

• “EFF Statement on California Governor’s Executive Order on AI”, by Rindala Alajaji — EFF Deeplinks, 2026-09-18T23:25:44.000Z (CC BY)
• Source URL: https://www.eff.org/deeplinks/2026/09/eff-statement-california-governors-executive-order-ai