(TNS) — San Francisco Mayor London Breed despatched a letter late Monday opposing state Sen. Scott Wiener’s landmark synthetic intelligence regulation invoice, the identical day tech billionaire Elon Musk got here out in assist of it.
In her letter, Breed stated whereas she agreed with the general intent of the laws, “extra work must be achieved to convey collectively business, authorities, and neighborhood stakeholders earlier than transferring ahead with a legislative answer that does not add pointless forms.”
Breed’s workplace couldn’t be reached to make clear what forms that referred to.
“With further time and collaboration, I’m assured that we are able to discover a answer that addresses most of the issues raised in latest months, whereas nonetheless enabling this rising subject to develop in a protected and sustainable approach,” Breed wrote, saying she opposed the invoice.
In an announcement, Wiener stated, “I totally assist Mayor Breed and her nice work to uplift San Francisco, however I respectfully disagree along with her opposition to SB1047 — a invoice that merely requires tech corporations to guage their largest and strongest AI fashions for danger of catastrophic hurt.”
“SB1047 is a pro-innovation, pro-safety invoice, which is why it is obtained such broad assist,” he added. “I stay up for persevering with to work with the Mayor on the huge array of points on which we’re deeply aligned, together with housing, leisure, and transportation. The wonderful thing about democracy is that we are able to comply with disagree on a difficulty whereas working collectively on different points. That is what we’ll do right here.”
Musk on Monday took a special strategy than some critics within the tech business, writing on X, “This can be a powerful name and can make some folks upset, however, all issues thought-about, I feel California ought to in all probability cross the SB1047 AI security invoice.”
Musk has repeatedly warned that runaway AI may pose a risk to humanity, and beforehand referred to as for a six-month pause on improvement of the know-how earlier than beginning his personal AI firm final yr.
“For over 20 years, I’ve been an advocate for AI regulation, simply as we regulate any product/know-how that could be a potential danger to the general public,” Musk stated.
The state invoice, which is up for an Meeting ground vote, has divided Democrats on the native and nationwide stage, drawing unusually broad consideration for a state measure.
Nancy Pelosi launched a letter opposing the invoice and a gaggle of Democratic members of Congress, together with Ro Khanna, D-Santa Clara, has additionally taken the admittedly uncommon step of opposing it in a letter this month.
The invoice would require the biggest and costliest AI fashions beneath the security testing to make sure they don’t seem to be able to inflicting catastrophic harm and lack of life, in addition to embrace “kill switches” to close down a program ought to it go awry. Opponents have stated the provisions are too broad and would discourage builders from constructing and utilizing AI techniques.
The invoice has drawn the ire of tech corporations and enterprise capitalists massive and small. Meta has opposed it, together with OpenAI and different makers of enormous AI applications. They argue it is going to quash innovation in a creating business key to the native and broader economic system.
Meta, which makes the open supply Llama household of free AI fashions extensively utilized by builders, has lobbied in opposition to the invoice, holding occasions and sending a letter to Wiener outlining why the corporate wouldn’t assist it. The co-founder of Stanford College’s Human-Centered AI Lab, Fei-Fei Li, has additionally come out in opposition to the invoice in a latest opinion piece and onstage at Stanford.
Not everybody in tech has come out in opposition to it. Main San Francisco AI developer Anthropic has recommended amendments, lots of which Wiener agreed to incorporate, and stated it may assist a model of the invoice.
These modifications embrace eradicating a provision that might see builders charged with perjury in the event that they failed to totally disclose the security testing they topic their AI applications to, and casting off the creation of a brand new company to supervise massive AI fashions on the state stage, as a substitute folding it into an current company.
The invoice has additionally obtained assist from some quarters of the tech neighborhood, together with former Google AI researcher Geoffrey Hinton and AI researcher Yoshua Bengio, and the Way forward for Life Institute, which warns in opposition to the catastrophic results of tech.
The Middle for AI Security Motion Fund, Financial Safety Motion California and Encode Justice have all signed on as co-sponsors of the invoice.
© 2024 the San Francisco Chronicle. Distributed by Tribune Content Agency, LLC.