How a California bill could have national implications in US’ tech race with China | Opinion
A California bill to regulate artificial intelligence fundamentally misunderstands the vector of AI progress and the high-stakes race between the West and China — a race that will determine which values dominate the 21st century.
Senate Bill 1047, the Safe and Secure Innovation for Frontier Artificial Intelligence Models Act, is authored by Sen. Scott Wiener, D-San Francisco. It would require developers to comply with new, strict requirements for the development of large AI models before training and deploying them. This AI regulation reflects a dangerous and naïve disconnect from the realities of technological advancement and geopolitical strategy.
Commercial interests who are behind in the race are egging on such action along with misinformed others. More advanced AI in China’s hands is a substantially larger risk to our country and our people than the small probabilities of sentient AI.
Opinion
Because of national security and global competitiveness implications, I believe the federal government should lead in regulating the safety of AI. Just last week, OpenAI formalized an agreement that allows the U.S. AI Safety Institute to receive access to major new models to inform safety best practices and standards. President Biden’s White House Executive Order was a well considered balance between safety monitoring and technical progress. The president’s approach had bipartisan support through the bipartisan Hill and Valley forum.
SB 1047, on the other hand, is built on the flawed premise that California regulations can contain AI development while still allowing the U.S. to maintain its technological edge. History has shown that such constraints are futile against a nation like China, which excels at bypassing unenforceable and one-sided regulations. When it comes to technologies like AI, which the Chinese Communist Party regards as a strategic asset worthy of wartime mobilization, no amount of legislation will compel China to play by the same rules.
China’s iron grip on TikTok’s algorithm — classifying it as a national security technology — is a clear indicator of how seriously they take AI as a tool of influence and in the techno-economic race with the west. The country that wins this technology race will win both the national security race, the cyber-offensive race and, more broadly, the race for technology to lead the future economy.
As a leading AI and climate investor, I’ve witnessed the exodus of climate technology operations due to California’s regulations. Our own green cement projects, initially based in California, have relocated to Utah because of regulations.
If this bill passes, companies will move critical projects out of California, leaving the state to lag behind in a race where it currently leads.
If this bill is enacted, California will become the last state to see the benefits of new AI technologies, receiving them only after they’ve been tested and vetted elsewhere. The open-source community, which thrives on unrestricted innovation, will suffer. But the biggest loser will be California, as it cedes its leadership to other states and countries.
Regulating AI at this stage is premature and misguided. By his own admission, Wiener wants to regulate “models that do not exist yet.” Would he have regulated smartphones in 2000? The constituencies pushing for these regulations are often the least equipped to understand the complexities of technological progress and the nature of iterative, punctuated, exponential progress.
Ian Stoica, a preeminent computer scientist at UC Berkeley, has warned that if this bill is signed into law, the frontier open source software used by Americans will come from “foreign nations” versus California. Is that Weiner’s idea of a safer world?
While the bill’s safety protocols are reasonable steps toward ensuring AI safety, there’s no need to throw the baby out with the bathwater. If those are truly the goals, we should pursue more surgical legislation that targets specific concerns without stifling broader innovation. Indeed, California should leverage its resources to fund safety research at state universities through initiatives like CalCompute. This would position the state as a leader in AI safety, rather than a roadblock to progress.
In the global race for AI supremacy, California is poised to be left behind if it continues down this path. The real threat isn’t the potential risks of AI itself, but the danger of ceding technological leadership to China, where AI is being developed without the constraints that this bill would impose. It’s time to prioritize strategic foresight over reactionary regulation, or we risk losing our edge in the most important race of our lifetime.
Vinod Khosla has been an investor for 40 years in California and was the co-founder of Sun Microsystems.