Debate in Silicon Valley ‘AI regulation law’ does not apply… Governor has veto power

Debate in Silicon Valley ‘AI regulation law’ does not apply… Governor has veto power


Refer Report

The governor of California vetoed an artificial intelligence (AI) regulation bill that caused great controversy, dividing the industry into pros and cons. If this bill, which shifts the focus of sanctions from ‘those who abused artificial intelligence models’ to ‘those who developed the model’, takes effect, major artificial intelligence companies such as OpenAI, Meta, Google, and Antropic, all headquartered in California, There was a lot of interest in whether or not the bill would go into effect, as these people were bound to be directly affected.

On the 29th (local time), California Governor Gavin Newsom, a Democrat, vetoed ‘SB 1047’, an artificial intelligence regulation bill that strengthens the responsibility of artificial intelligence development companies. The deadline for signing the veto was the 30th. The Democratic-dominated state Senate and House of Representatives passed this bill almost unanimously on the 28th of last month.

Governor Newsom“The core of the debate is whether to set regulatory standards based on the cost and amount of computation required to develop artificial intelligence models,” he said. “This bill focuses only on large-scale models with the most expensive training costs.” “A careful balance is needed as small, specialized models that may be more risky may emerge.” He said, “I have the same opinion as the bill sponsor. Safety protocols must be adopted and proactive safeguards must be introduced. But it must be based on empirical evidence and science,” he said.

Los Angeles Times“It is possible, but extremely rare, for a bill to override a governor’s veto with a two-thirds vote in both chambers,” he said. Governor Newsom said he would continue to cooperate with the legislature regarding artificial intelligence regulation and hinted at the possibility of revising the bill.

The bill mandates safety testing only for large-scale language models that have spent more than $100 million (132.7 billion won) on training, and requires developers to take reasonable care to ensure that artificial intelligence systems do not cause catastrophic harm, such as multiple deaths or property damage of more than $500 million. It is stipulated that: If fatal damage occurs, a fine of 10% of the model training cost must be paid.

Immediately after the proposal last February, there was strong opposition from the tech industry, including Microsoft (MS), Meta, and Open AI. Leading Democratic figures, including former House Speaker Nancy Pelosi and San Francisco Mayor London Breed, also publicly opposed it. However, technology experts and artificial intelligence experts, including Elon Musk, supported the bill, leading to a long controversy in the tech industry and academia.

Reporter Kim Won-cheol wonchul@hani.co.kr

Source: Korean