California’s proposal to impose a new tax on large tech companies has faced significant opposition from numerous organizations in Silicon Valley. California lawmakers narrowly acquiesced to pressure, incorporating several amendments recommended by AI firm Anthropic and other detractors.
California’s Appropriations Committee passed an invoice on Thursday, a crucial step towards becoming law, according to Senator Wiener’s office, which notes several key amendments were made.
The senator declared, “We’ve successfully integrated numerous cost-effective modifications, effectively addressing the primary concerns raised by Anthropic and other industry stakeholders, as I see it.” “These revisions build upon significant changes I previously introduced to Senate Bill 1047, carefully tailored to meet the unique needs of the open-source community – a vital source of innovation.”
While SB 1047 aims to prevent massive AI programs from causing catastrophic harm, including the loss of multiple lives and cybersecurity incidents exceeding $500 million in damages, by imposing liability on developers. Despite this development, the revised invoice significantly curtails California’s regulatory powers to scrutinize AI laboratories’ activities.
Introduced by Senator [name] in the [legislative session], Senate Bill 1047 aims to [briefly describe the bill’s purpose]. Specifically, this legislation seeks to [list specific provisions or changes made by the bill]. If enacted, SB 1047 would have a significant impact on [affected area or group], potentially leading to [expected outcomes or consequences].
While California’s laws allow professional norms to hold AI companies accountable for lax security measures, the current framework does not permit lawsuits prior to a devastating event happening. This proposal comes from Anthropic.
California’s legal authorities can seek injunctive relief by requesting that a company cease a specific operation deemed detrimental, and may still pursue litigation against an AI developer if their model inadvertently triggers a catastrophic event.
Additionally, Senate Bill 1047 does not establish the Frontier Manikin Division (FMD), a novel administrative entity created within the legislation as originally proposed. Notwithstanding its shortcomings, the proposed invoice would establish the Board of Frontier Fashions as the foundation of the Fashion Management Department (FMD) within the existing Government Operations Agency. The board has expanded to accommodate nine members, replacing its original composition of five individuals. The Board of Frontier Fashions will establish computerised thresholds for linings, adhering to security guidance and applicable laws governing auditors’ conduct.
Senator Wiener modified SB 1047 to exempt AI labs from submitting security clearance results under penalty of perjury, instead mandating publicly disclosed statements outlining their security protocols with no enforceable consequences.
SB 1047 has been updated to include more accommodating terminology regarding how developers ensure AI models are secure. The revised text reads: Now, the invoice necessitates that builders provide “affordable care” options, not posing a significant threat of causing catastrophe, in place of the “affordable assurance” previously mandated by the invoice.
Lawmakers have introduced measures to ensure the security of open-source fashion trends. According to SB 1047, an individual who expends less than $10 million customizing a life-sized mannequin is not considered a developer. The responsibility will ultimately rest with the primary developer of the prototype.
What’s driving these sudden changes in our workflow processes?
While the proposed invoice has faced significant resistance in the United States, Congressmen, renowned AI researchers, large tech companies, and venture capitalists, the bill has flown through California’s legislature with relative ease. To alleviate concerns surrounding SB 1047, these revisions aim to craft a more palatable bill that satisfies both opponents of the original legislation and Governor Newsom, ultimately securing his signature without compromising support from the AI industry.
While Governor Newsom has remained silent on SB 1047, his commitment to advancing California’s artificial intelligence sector remains unwavering.
Those proposed changes will hardly satisfy the most ardent detractors of SB 1047. While the revised bill may lack the initial strength of its predecessors, it still obliges developers to account for the liabilities arising from their artificial intelligence models. The assumption that Senate Bill 1047’s fundamental tenets should be uniformly endorsed is misguided, and the proposed amendments fail to adequately address this concern.
Immediately following the passage of SB 1047 on Thursday, eight U.S. representatives from California penned a letter to Governor Newsom urging him to exercise his veto power and reject the bill. The government’s decision to draft an invoice that restricts the development of artificial intelligence “would not be beneficial for our state, the startup ecosystem, scientific progress, or national defence, which could potentially suffer harm as a result of unchecked AI advancement.”
What’s subsequent?
SB 1047 is slated to reach California’s Senate floor for its final vote. Since the bill has passed, it will need to be referred back to California’s Senate for reconsideration and another potential vote due to the newly introduced amendments. If passed, the bill would then proceed to Governor Newsom’s office, where he may choose to either veto it or sign it into law.
