Proposed law to control powerful AI models will destroy California’s nascent industry

A new bill introduced in the California State Senate threatens to crush the state’s vibrant artificial intelligence industry under the weight of draconian regulations and government overreach.

Senate Bill 1047, deceptively titled the “Safe and Secure Innovation for Frontier Artificial Intelligence Models Act”, would impose onerous restrictions and oversight on any company developing cutting-edge AI models.

If passed into law, it will kill innovation, concentrate power in the hands of tech giants, and cost California thousands of jobs – all while doing virtually nothing to actually protect consumers. Here’s why.

Vague definitions and lack of focus on consumer harms

The bill’s vague language gives nearly limitless power to a newly created government bureaucracy, the so-called “Frontier Model Division”, to define and police supposed “hazardous capabilities” of AI models. Startups and researchers will be forced to beg government apparatchiks for permission to innovate. The compliance burdens and legal risks will make it impossible for any but the largest tech companies to compete in AI development. Smaller players without armies of lawyers will be regulated out of existence.

Meanwhile, the bill contains no concrete provisions to prevent specific harms to consumers.

Its hand-waving appeals to nebulous concepts like “public safety” and “responsible practices” are a smokescreen for the real purpose – expanding government control over one of the most exciting and dynamic sectors of our economy.

This is the epitome of nanny-state paternalism. Do California legislators really believe they should dictate the future of one of the 21st century’s most important technologies?

First Amendment violation

SB 1047 is also a slap in the face to fundamental American values like free speech and free enterprise. Software code has long been recognized as a form of expressive speech protected by the First Amendment. By subjecting AI code to prior restraint and post-hoc penalties, this bill shreds critical Constitutional safeguards and sets a dangerous precedent for government censorship of creative expression.

The supposed public benefits of SB 1047 are speculative at best, but the damage it will cause is all too predictable. Investment capital and entrepreneurial talent will flee California for jurisdictions that don’t treat AI innovation like a crime. Instead of securing America’s place as the global leader in artificial intelligence, California will drive this transformative technology straight into the welcoming arms of our geopolitical rivals.

California has long prided itself on being a hub for dreamers, disruptors, and technological visionaries. But if SB 1047 becomes law, that dream will turn into a nightmare of Big Government strangulation. This bill must be opposed vigorously by anyone who values American dynamism and ingenuity. The future of AI should be built through the creative genius of free minds, not the dead hand of government bureaucrats.

A recipe for stifling innovation

A close look at the bill’s provisions reveals just how destructive SB 1047 would be for AI innovation in California. The heart of the bill is its requirement that developers of “covered models” – defined so broadly it could encompass virtually any significant AI system – submit to a draconian regulatory regime. This includes invasive annual certifications, mandatory implementation of ill-defined “safety protocols,” and giving the government intrusive access to confidential information under threat of crippling penalties.

SB 1047 also empowers unaccountable bureaucrats to define and ban any AI capabilities they unilaterally deem “hazardous,” with violators facing company-ending fines of up to 30% of development costs. This vague and arbitrary power would create paralyzing uncertainty, as businesses could never be sure when regulators might decide to destroy their investments on a politicized whim.

Government controlled AI kill switches?

Most chillingly, SB 1047 requires AI developers to implement “full shutdown” capabilities, giving government officials a kill switch to unilaterally terminate any AI system they disapprove of. This is a recipe for abuse and a grievous assault on the free speech rights of software creators. The power to compel any company to destroy its own products and property at the government’s command is the power to hold every business hostage to bureaucratic extortion.

These heavy-handed provisions impose immense compliance costs that would make it nearly impossible for startups or small businesses to compete in AI development. Only the largest tech giants with armies of lawyers could navigate the bill’s legal minefield and absorb its regulatory taxes. Far from protecting consumers, SB 1047 entrenches tech monopolies by raising insurmountable barriers for would-be competitors.

By inserting government control into the heart of the creative process, SB 1047 would cripple the cycles of dynamic experimentation and iterative advancement that drive progress in AI. Companies would be forced to shape their innovations to appease regulators rather than serve users. Research areas that bureaucrats frowned upon would be abandoned. California would see a devastating brain drain as entrepreneurs and engineers flee to states that don’t criminalize AI invention.

Bureaucrats playing God with the future

The hubris of SB 1047’s authors is matched only by their naivete about the realities of AI development. The bill is premised on the fantastical belief that a handful of government regulators can outthink the combined wisdom of the world’s brightest AI researchers and entrepreneurs. But the uncomfortable truth is that the sorts of people who dedicate their careers to pushing the boundaries of human knowledge in frontier industries rarely end up working in government bureaucracies.

It strains credulity to imagine that the Frontier Model Division will be staffed with the world-class AI experts needed to make wise decisions about mind-bogglingly complex technologies. More likely, it will be populated by well-meaning but overmatched regulators who lack the technical sophistication to distinguish truly hazardous AI from exciting innovations that should be encouraged.

By centralizing control over AI in the hands of those furthest from the cutting edge, SB 1047 virtually guarantees a raft of destructive unintended consequences. Regulators will be perpetually behind the curve, fighting yesterday’s battles while hampering the development of tomorrow’s breakthroughs. Entire categories of research will be stifled as bureaucratic gatekeepers block any work they don’t understand or that clashes with political agendas.

This isn’t just a threat to individual companies, but to the future health of California’s entire economy. Throughout history, the advent of transformative technologies like the steam engine, electricity, and the internet have unleashed long waves of creative destruction and growth. Economists like Carlota Perez have shown how these technological revolutions drive 40-60 year cycles of disruptive innovation and productivity growth. Artificial intelligence has the potential to be the defining technology of the next wave – but only if its development is not prematurely cut off by misguided regulations like SB 1047.

By strangling AI in its crib, California risks squandering a once-in-a-generation opportunity to ride the crest of the coming wave of innovation. The jobs, companies, and industries that will define the economy of the 2040s and 2050s will be the ones that harness machine learning to revolutionize their fields. If California succeeds in driving this revolution beyond its borders, it will have only itself to blame when those opportunities enrich its rivals instead.

California’s Choice: Lead the AI revolution or fade to irrelevance 

If the authors of SB 1047 were truly motivated by a desire to protect consumers and promote responsible AI development, they would focus on targeted regulations that directly address real harms. Instead of vague and innovation-crushing restrictions, they could craft legislation to safeguard users’ data privacy, create a bill of rights enshrining individuals’ ownership and control over their personal information, prevent AI from being weaponized to make online experiences addictive and exploitative, and defend the intellectual property rights of content creators whose work is currently appropriated without compensation to train AI models.

But SB 1047 does none of this. Its provisions are either so broad as to be meaningless, or so burdensome as to make AI development untenable for any but the largest tech giants. By fixating on nebulous “hazardous capabilities” while ignoring concrete abuses, the bill reveals its sponsors’ true priority: expanding government control over a vibrant and strategically vital industry.

The folly of this approach is already evident in the exodus of major tech companies from California in recent years. Oracle, Hewlett Packard Enterprise, Tesla, Dropbox, and Palantir Technologies have all moved their headquarters out of state, driven away by the hostile business climate created by excessive regulation and taxation. If SB 1047 becomes law, it will only accelerate this brain drain, as AI startups and researchers flee to jurisdictions that don’t treat them like enemies of the state.

Californians deserve better than this innovation-killing power grab. State leaders should be working to foster a thriving AI ecosystem that balances exciting technological progress with thoughtful protections for individuals’ rights and creative work. With more carefully tailored rules, California can remain a global center for turning cutting-edge research into transformative products and services that enrich people’s lives.

SB 1047 threatens to quash that bright future in a tangle of red tape and bureaucratic overreach. It must not be allowed to become law. California’s destiny should be as a launchpad for the revolutionary innovations of the 21st century, not a cemetery where visionary ideas go to die at the hands of shortsighted regulators. State legislators must reject this bill and pursue an AI policy agenda worthy of the state’s innovative spirit – one that supports responsible technological breakthroughs, not heavy-handed government meddling in a field bureaucrats are ill-equipped to micromanage.

The choice is clear: California can steer the AI revolution from the front, or it can squander its tech leadership and watch this transformative technology blossom elsewhere. For the sake of the state’s economic vitality, entrepreneurial dynamism, and global competitiveness, there can be only one answer. SB 1047 must be defeated, and California must recommit to stewarding AI’s epochal potential for empowering progress. The alternative is to continue driving away the innovators who have made the state a global tech powerhouse, and consign California to a future of stagnation and decline.