Elon Musk and Vitalik Buterin have expressed their support for the widely debated California bill SB 1047, aimed at regulating artificial intelligence.
This is a tough call and will make some people upset, but, all things considered, I think California should probably pass the SB 1047 AI safety bill.
For over 20 years, I have been an advocate for AI regulation, just as we regulate any product/technology that is a potential risk…
— Elon Musk (@elonmusk) August 26, 2024
“For over 20 years, I have been an advocate for AI regulation, just as we regulate any product/technology that poses a potential risk to society,” noted the billionaire.
In February, the “Safe and Secure Innovations for Advanced AI Models” bill, known as SB 1047, was proposed. If passed, developers will be required to implement safety protocols to prevent mass casualties or major cyberattacks, as well as introduce an “emergency stop” feature for AI models.
The bill also mandates annual independent safety audits for AI. The rules will apply only to the largest models, which cost at least $100 million and use 10^26 FLOPS for training. They will primarily affect giants like OpenAI, Google, and Microsoft.
Later, California legislators introduced several amendments, mainly aimed at reducing the state government’s authority to hold AI labs accountable. However, the document still holds developers responsible for the dangers of their models.
Vitalik Buterin commented on Elon Musk’s post, noting that the document’s goal is to make safety testing mandatory before launching artificial intelligence that “could be dangerous to the world.”
What’s the best evidence that the bill is going to be used to go after open weights?
I know that earlier versions of the bill had a full shutdown req that’s incompatible with open weights, but that’s been removed. And I know that some AI safety people have expressed support for…
— vitalik.eth (@VitalikButerin) August 27, 2024
Major AI startup Anthropic supported SB 1047, noting that “the benefits of the bill likely outweigh the costs.” Amendments to the original document were made at the company’s request.
Criticism
OpenAI opposed SB 1047, arguing that the bill would drive talent out of California and stifle innovation.
“The AI revolution is just beginning, and California’s unique status as a global leader in this field underpins the state’s economic dynamism. SB 1047 will jeopardize this growth, slow the pace of innovation, and force world-class California developers and entrepreneurs to leave the state in search of new opportunities elsewhere,” emphasized OpenAI’s Chief Strategy Officer Jason Kwon.
Founder of the startup Extropic and former Google developer Guillaume Verdon noted that “AI can be powerful, but an authoritarian government with shadowy control over powerful AI is scarier.”
“This sets a terrible precedent at the national level and will open the door for a legal war by the government against companies that do not offer it ‘backdoors’ in LLM,” added the expert.
He compared the situation to the recent arrest of Telegram founder Pavel Durov and the implications for freedom of speech and decentralization.
The document also faced criticism from Google, Meta, renowned experts, and others. Zuckerberg’s company considered that the bill threatens to make the region unfavorable for AI development and deployment.
When Will It Be Passed?
In mid-August, the bill was backed by the California Appropriations Committee. It has been sent to the California Assembly floor for a final vote. If this stage is passed, a vote in the state Senate will follow, after which the governor’s signature is required, who may veto it.
In August, a comprehensive law to regulate AI came into force in the EU.