California is going on the offensive when it comes to AI regulation. On Monday, March 30, 2026, Democratic Governor Gavin Newsom signed an executive order (N-5-26) that places stricter requirements on AI companies that want to do business with the state. In doing so, California is deliberately setting itself apart from the Trump administration, which is aiming to deregulate the AI sector as far as possible.
California is regarded as a global center of AI innovation – according to the decree, 33 of the world’s 50 leading AI companies are based in the state. The signal effect of this step is correspondingly great.
What the new executive order specifically provides for
Essentially, the executive order gives the Californian authorities 120 days to develop new certification and safety standards for AI companies. In future, companies wishing to conclude contracts with the state will have to prove that they have effective protective measures in place. This includes the following areas, among others:
- Prevention of illegal content: This includes, in particular, the dissemination of child sexual abuse material (CSAM) and non-consensual intimate recordings.
- Reduction of AI bias: Companies must explain how they recognize and reduce harmful biases in their models.
- Protection of civil rights: Protection against unlawful discrimination, surveillance and imprisonment is explicitly demanded.
- Labeling of AI content: Authorities should develop procedures to watermark AI-generated or manipulated images and videos – an important step against deepfakes and disinformation.
The decree also provides for the federal state to expand its own use of generative AI. Among other things, there are plans for an AI-supported portal that citizens can use to find government services according to life events such as starting a business or looking for a job.
Direct confrontation with Washington
The timing and wording of the decree leave no doubt that Newsom deliberately wants to set a counterpoint to the Trump administration. It was only in December 2025 that Trump issued his own decree in which he formulated a national framework for AI – and in which he virtually called on the federal states to dispense with their own regulations. The tenor of the Trump decree was that innovation needs freedom, not onerous regulations.
In addition, the US Department of Justice set up its own task force in January 2026 to legally challenge regulatory attempts by individual states in the field of AI.
Newsom opposes this. In a statement, he explained that California, as the birthplace of the tech industry, will ensure that AI companies protect the rights of citizens – and do not exploit them.
The Anthropic case also plays a role
The executive order comes at a time when tensions between the Trump administration and the AI industry are increasing anyway. Only recently, the Pentagon classified the AI company Anthropic as a supply chain risk and banned government contractors from using its technology. However, a US federal court temporarily suspended this decision by means of a temporary injunction, describing the classification as possibly unlawful and politically motivated.
The decree also stipulates that California may independently review federal security classifications. If the state’s Chief Information Security Officer concludes that a classification is unjustified, Californian authorities can continue to cooperate with the company concerned.
Conclusion: California as a blueprint for AI regulation?
With the new Executive Order, California is positioning itself as a pioneer in AI regulation – not just in the USA, but worldwide. The approach of enforcing safety standards by awarding public contracts is pragmatic and can work faster than lengthy legislative procedures. Whether Trump’s task force will take action against this remains to be seen. One thing is clear: the power struggle over the rules of the game in artificial intelligence has only just begun in the USA.
