[AI Hanip News] The Strongest Regulation Is Coming... What Is 'SB1047'?

AI Regulation Passed by California Legislature
Mandatory AI Pre-Testing and Kill Switch
Silicon Valley Big Tech Faces Urgent Threat

Silicon Valley, the global technology hub including artificial intelligence (AI), is trembling before the strongest AI regulations. This is because California, the state where Silicon Valley is located, has passed the 'Safety and Security Innovation Act for Advanced AI Models (SB1047)'.

On the 29th (local time), SB1047 passed the California State Senate. This followed its approval in the Assembly with 41 votes in favor and 9 against.

The bill mandates companies developing AI models to undergo safety testing before disclosing their technology. It requires pre-verification to ensure AI does not engage in dangerous activities such as developing hazardous weapons. Additionally, developers must include a 'Kill Switch' function to shut down the AI model if it causes problems. Measures must also be put in place to protect internal whistleblowers reporting AI issues. Furthermore, the state Attorney General can sue developers if AI causes human fatalities or property damage amounting to $500 million (approximately 670 billion KRW).

Not all AI companies are immediately subject to this. Only AI models with development costs exceeding $100 million (approximately 130 billion KRW) or requiring a certain scale of electricity consumption are mandated to undergo pre-testing. Although this depends on specific calculation criteria, most big tech companies that have developed large language models (LLMs) are expected to be affected.

Previously, the European Union (EU) became the first in the world to pass a comprehensive AI law regulating AI companies and technologies. AI related to public services and national systems such as healthcare, education, and finance is classified as 'high-risk' and is subject to supervisory oversight. This appears to be an effort to secure leadership in AI regulation and narrow the technological gap with the United States. It is notable that the U.S., which has led AI technology, and specifically Silicon Valley, the city of innovation, is now moving toward regulation. Although other U.S. states like New York have pursued or introduced AI regulatory bills, none have been as stringent.

Opinions in Silicon Valley are sharply divided. OpenAI opposed the bill, warning that companies might leave California if it passes. The industry group 'Chamber of Progress', representing Google, Apple, Amazon, and others, also demanded the governor veto the bill, arguing that it should not suppress key sectors of California's economy. They expressed concerns that excessive regulation would hinder technological development and innovation.

On the other hand, xAI, founded by Elon Musk, and Anthropic, considered a rival to OpenAI, support the bill. Elon Musk, CEO of Tesla, endorsed the bill on his 'X' account, calling it "regulating all products and technologies that pose potential public risks." Musk, a latecomer in the AI field, has previously warned about the dangers of AI technology and sided with regulators. Anthropic, advocating for safe AI, stated in an open letter that "the bill addresses catastrophic risks of AI" and that "small companies are excluded from the scope, greatly reducing concerns about stifling innovation."

The remaining step is the signature of California Governor Gavin Newsom. Governor Newsom must decide whether to sign the bill by the 30th of next month. Given the sensitivity of the issue, he has not yet expressed a position. If he signs the bill, major AI companies will effectively fall under its influence, drawing significant attention to the repercussions.