加州州長 Gavin Newsom 否決人工智慧安全法案 SB 1047，理由是對創新進行不必要的限制

埃里溫 (CoinChapter.com) — 9 月 30 日，加州州長 Gavin Newsom 否決了 SB 1047，一項廣泛討論的人工智慧安全法案。

California Governor Gavin Newsom has vetoed a bill that aimed to regulate artificial intelligence (AI) safety in the state, following strong opposition from major tech companies.

在大型科技公司的強烈反對下，加州州長加文·紐瑟姆 (Gavin Newsom) 否決了一項旨在監管該州人工智慧 (AI) 安全的法案。

The bill, SB 1047, was titled the Safe and Secure Innovation for Frontier Artificial Intelligence Models Act. It faced significant opposition from the tech industry in Silicon Valley, which argued that the bill's regulations would hinder AI development in California and ultimately harm innovation.

該法案名為 SB 1047，名為《前沿人工智慧模型安全創新法案》。它遭到了矽谷科技業的強烈反對，他們認為該法案的規定將阻礙加州人工智慧的發展，最終損害創新。

In a statement on Sep. 29, Newsom said that the bill had good intentions but focused too heavily on regulating existing AI firms. He added that the bill failed to address the real risks posed by evolving AI technologies and instead imposed unnecessary restrictions on even basic AI functions.

紐瑟姆在 9 月 29 日的聲明中表示，該法案的初衷是好的，但過於關注對現有人工智慧公司的監管。他補充說，該法案未能解決不斷發展的人工智慧技術帶來的真正風險，反而對基本的人工智慧功能施加了不必要的限制。

"This bill largely applies retroactively to a small number of existing technologies and does not advance our goal of ensuring the safe development of future AI models," Newsom said in his veto message.

紐瑟姆在否決資訊中表示：“該法案在很大程度上追溯適用於少數現有技術，並且不會推進我們確保未來人工智慧模型安全開發的目標。”

The bill proposed mandatory safety tests for AI models and required developers to implement a "kill switch" for their AI systems, which could be used by the state to deactivate the models in case of extreme risks. It also stated that AI developers could face legal action from the state attorney general if their models posed a threat, such as taking control of critical infrastructure.

該法案提出對人工智慧模型進行強制性安全測試，並要求開發人員為其人工智慧系統實施“終止開關”，國家可以利用該開關在出現極端風險時停用模型。它還表示，如果人工智慧開發人員的模型構成威脅，例如控制關鍵基礎設施，人工智慧開發人員可能會面臨州總檢察長的法律訴訟。

The bill targeted large-scale AI models that cost over $100 million to develop, which raised concerns from industry experts. Neil Chilson, head of AI policy at the Abundance Institute, told CoinChapter that the bill's scope could eventually expand to cover smaller developers as well.

該法案針對的是開發成本超過 1 億美元的大規模人工智慧模型，這引起了行業專家的擔憂。 Abundance Institute 人工智慧政策負責人 Neil Chilson 告訴 CoinChapter，該法案的範圍最終可能會擴大到涵蓋小型開發人員。

"The bill is poorly drafted and could be interpreted to apply to any company using AI models, even small startups," Chilson said.

奇爾森說：“該法案起草得不好，可以被解釋為適用於任何使用人工智慧模型的公司，甚至是小型新創公司。”

Governor Newsom mentioned that he had consulted with AI safety experts and emphasized the need for "workable guardrails" based on scientific analysis. He ordered state agencies to continue evaluating the potential risks posed by advanced AI models.

紐瑟姆州長提到，他已經諮詢了人工智慧安全專家，並強調需要基於科學分析的「可行的護欄」。他命令國家機構繼續評估先進人工智慧模型帶來的潛在風險。

"While I am vetoing this bill today, I want to be clear that I fully support the Legislature's efforts to evaluate the risks and benefits of advanced AI models and to put in place strong safety measures," Newsom said.

紐瑟姆說：“雖然我今天否決了這項法案，但我想明確表示，我完全支持立法機關評估先進人工智能模型的風險和收益並採取強有力的安全措施的努力。”

Several lawmakers, including former House Speaker Nancy Pelosi, opposed SB 1047, arguing that the bill would hinder AI innovation, which is crucial to California's economy. Major tech firms also expressed concerns that the bill would limit AI development and growth.

包括前眾議院議長南希·佩洛西在內的幾位議員反對 SB 1047，認為該法案將阻礙人工智慧創新，而人工智慧創新對加州經濟至關重要。大型科技公司也擔心該法案將限制人工智慧的發展和成長。

In contrast, Elon Musk supported the bill and broader AI regulations. On Aug. 26, Musk posted on X, saying that

相比之下，馬斯克支持該法案和更廣泛的人工智慧法規。 8月26日，馬斯克在X上發文表示，

"California should probably pass the SB 1047 AI safety bill,"

“加州或許應該通過 SB 1047 人工智慧安全法案，”

though he acknowledged that it was a difficult decision.

儘管他承認這是一個艱難的決定。

Despite his veto, Newsom stressed the importance of adopting strong AI safety measures, citing that his administration had passed over 18 AI-related bills in the past 30 days.

儘管擁有否決權，紐瑟姆還是強調了採取強有力的人工智慧安全措施的重要性，並指出他的政府在過去 30 天內通過了超過 18 項與人工智慧相關的法案。