The hottest artificial intelligence (AI) innovation news of 2025 is coming out of China, with cutting-edge products like DeepSeek R1 and Manus creating a modern “Sputnik moment” and “wake-up call” for the United States, which was previously thought to be well ahead of the world. This means the governance choices made next will have profound ramifications for American innovation, competitiveness and geopolitical security.

The United States stands at a crucial crossroads with two different AI policy paths to choose from. One leads to the freedom to innovate and freedom of speech. The other leads to European-style regulatory repression of commerce and content.

Unfortunately, federal and state lawmakers are headed in different directions. While the Trump administration has been taking steps to embrace a pro-freedom AI opportunity and growth agenda, many states are instead giving Europe’s regulatory model serious consideration.

According to one AI bill tracking service, less than three months into 2025, nearly 900 AI-related measures are already moving, exceeding the total number of AI bills introduced in all of 2024. Most are state bills that would expand government control over algorithmic systems in some fashion.

This unprecedented volume of regulatory activity threatens to undermine AI innovation and investment in America through proliferating parochial paperwork hassles and compounding compliance costs. Many of these state proposals essentially follow the European Union’s (EU) regulatory playbook with laws that mimic the EU’s new AI Act.

That’s a recipe for technological stagnation, as Europe’s experience proves. Thirty years of top-down over-regulation left Europe suffering from a continent-wide competitiveness crisis and no major global digital technology leaders. The EU’s primary digital exports today are cookie pop-ups and compliance forms, not world-class companies or products.

Why would states want to import these failed policies to America?

Colorado already passed one such Euro-style AI law last May, although Gov. Jared Polis (D) admitted it would create “a complex compliance regime for all developers and deployers of AI,” and will “tamper innovation and deter competition.” Unfortunately, the Colorado law and dozens of copycat “algorithmic discrimination” bills combine the worst elements of EU techno-planning with the Biden administration’s heavy-handed AI vision, which President Trump has since wisely jettisoned.

These state bills look to create the equivalent of AI pre-crime divisions, with bureaucrats empowered to preemptively sniff out examples of potentially unfair or “biased” AI-generated outputs for applications that have not yet even been deployed. “Algorithmic impact assessments” will be required for “consequential decisions” to ensure AI systems are “in the public interest.”

Imagine if similar vague, technocratic mandates had been imposed by 50 different state internet bureaus at the dawn of the personal computing age. America’s digital technology revolution would have fizzled. Luckily, in the mid-1990s, policymakers instead opted for a pro-freedom paradigm for internet and digital commerce, and America experienced an outpouring of innovation and speech opportunities. The key to this was a policy regime that protected the digital marketplace from state and local over-regulation. 

Similarly, Congress should comprehensively preempt state and local AI regulations that impinge upon interstate algorithmic commerce and speech. Even Colorado’s Polis has called for Congress to craft “a needed cohesive federal approach . . . to limit and preempt varied compliance burdens on innovators and ensure a level playing field across state lines.”

Short of full-blown preemption, federal lawmakers could craft an “AI learning period moratorium” that would limit new federal and state AI regulatory mandates that undermine a competitive national marketplace. This would give AI entrepreneurs some breathing room to launch bold new ideas and products to meet the rising global competition, while policymakers study which policies make sense. 

Some issues or sectors covered by traditional state authority will need to be carved out, including education and law enforcement. But Congress can take steps to clarify what AI models and applications are covered by federal laws.

State lawmakers who want to proactively protect AI opportunity in America should embrace policies that protect the freedom to code, compute and innovate. So-called “right to compute” measures are moving in Montana and New Hampshire that would protect the right of individuals to access and use computational systems, treating over-regulation as an infringement of citizens’ fundamental rights to property and free expression.

Utah has also passed an AI law that creates a “Learning Laboratory Program” to encourage innovators to work together with state officials to foster new AI applications and study flexible approaches for sectors already regulated. This approach represents the time-tested and uniquely American innovation policy paradigm that is more agile, incremental and prudent as compared to Europe’s.

America’s AI opportunity agenda must embrace freedom and reject fear to give our nation’s brilliant innovators a green light to once again develop life-enriching technologies that lead the world. But state regulators must get out of the way.

Adam Thierer is a resident senior fellow with the R Street Institute’s technology and innovation team