California Governor Gavin Newsom issued an executive order Monday that will require artificial intelligence (AI) companies seeking state contracts to demonstrate robust safety and privacy standards. This move marks a significant step in state-level AI regulation, setting a new precedent for how governments interact with rapidly evolving AI technologies.
Vetting AI Contractors
Under the new rules, companies competing for state business must disclose their AI safety and privacy policies upfront. California will scrutinize these policies to ensure they actively prevent exploitation, including the distribution of illegal content like child sexual abuse materials. The state will also assess whether AI systems are used for unwarranted surveillance or censorship, and whether developers are taking steps to mitigate bias in their algorithms.
This isn’t just about ticking boxes. The vetting process is critical because AI systems are increasingly deployed in sensitive areas, and unchecked biases or privacy violations can have far-reaching consequences. Without transparency, the risk of misuse grows substantially.
Independence from Federal Standards
California will not automatically defer to federal assessments of AI companies. Even if the Pentagon designates a firm as a supply chain risk (as recently happened with AI startup Anthropic), the state will conduct its own independent evaluation. This move signals California’s willingness to forge its own path in AI oversight, even when it diverges from federal policy.
The Pentagon’s dispute with Anthropic is a case in point. The Defense Department severed ties with the AI company after Anthropic refused to allow the use of its models for mass domestic surveillance or autonomous weapons deployment. This underscores a fundamental tension between aggressive military applications of AI and ethical considerations.
Watermarking AI-Generated Content
The order also directs state agencies to watermark any AI-generated or manipulated videos they produce. This measure is aimed at combating misinformation by making it easier for the public to distinguish between authentic and AI-created content. By labeling state-produced AI content, California acknowledges the growing threat of deepfakes and synthetic media.
This is a proactive step towards building trust in digital media. The rise of AI-generated imagery means consumers need reliable tools to verify authenticity, and watermarking is one such tool.
California’s executive order is more than just a bureaucratic change; it’s a statement that AI innovation must be responsible. The state is signaling its intention to set the terms for how AI operates within its borders, prioritizing safety, privacy, and transparency over unchecked deployment.
