The AI Daily Brief: Artificial Intelligence News and Analysis - Newsom Vetoes SB-1047, Sending Controversial AI Regulation Fight National

Episode Date: October 1, 2024

Governor Gavin Newsom has vetoed SB-1047, a major bill that aimed to regulate AI development in California by imposing safety standards and liability for developers. The decision has sparked both disappointment and celebration across the tech world, with proponents calling it a missed opportunity for oversight, while critics argue it would have stifled innovation. The debate is set to move to the national stage—what’s next for AI regulation in the US? Learn how to use AI with the world's biggest library of fun and useful tutorials: https://besuper.ai/ Use code 'youtube' for 50% off your first month. Concerned about being spied on? Tired of censored responses? AI Daily Brief listeners receive a 20% discount on Venice Pro. Visit ⁠⁠⁠https://venice.ai/nlw⁠⁠⁠ and enter the discount code NLWDAILYBRIEF. The AI Daily Brief helps you understand the most important news and discussions in AI. Subscribe to the podcast version of The AI Daily Brief wherever you listen: https://pod.link/1680633614 Subscribe to the newsletter: https://aidailybrief.beehiiv.com/ Join our Discord: https://bit.ly/aibreakdown

Today on the AI Daily Brief, SB 1047 has officially been vetoed. Before that in the headlines, Microsoft is finally releasing a very controversial AI feature that was announced earlier this year. The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. To join the conversation, follow the Discord link in our show notes. Welcome back to the AI Daily Brief Headlines edition, all the daily AI news you need in around five minutes. We kick off today with the return of a feature, which on the one hand was very controversial. upon its announcement, but on the other hand, feels absolutely inevitable in the future that we are heading into. That is Microsoft's recall feature. Basically, the idea of recall and the tools like

it is that by watching what a user is doing and feeding that information into AI, a computing experience can be much more responsive to a user's actual needs. Now, Microsoft is, of course, not the first company who's trying something like this. Limitless has been working on this, for example, for quite some time now. First, as a desk, and web app, and more recently coming to a wearable format as well. The way that they frame it is going beyond your mind's limitations. This is effectively the Penceive from Harry Potter, which allows you to externalize all of these memories so that you can go back in and recall them with the help of AI whenever you need. While Microsoft announced earlier this year that it was

bringing that type of experience to their co-pilot plus computers, but after an uproar from privacy advocates, they paused the rollout for a little while. Microsoft now says that it has quote, listened to feedback and has made some updates, for example, making the service opt-in, whereas previously in these computers it had been turned on by default. In the wake of the announcement, it wasn't just online advocates who were concerned. There were government bodies as well, such as the Information Commissioner's Office in the UK, who looked into the system and were influential in some of these new updates. For my part, I completely understand why this is controversial. I have a base level, nervousness and skepticism when it comes to anything

with this level of surveillance. And yet, at the same time, I'm fairly sure that this is going to be totally day-reguer within the next, I don't know if it's five or ten years, but at some point, this will just be completely normalized. I think the value propositions will be there, particularly from a business standpoint, and as is always the case, people will just get used to something that a previous generation never would have countenanced. I could, of course, be wrong, and I think this is a good one to keep an eye on for what sort of norms we want as it relates to this in society, but my guess is you're going to see more, not less of this, in the years to come. Moving on, a follow-up from our story last week about OpenAI's big investment round.

Apple apparently is not going to be joining that round. Now, according to the Wall Street Journal, this $6.5 billion mega round could close as early as this week. It's not clear why Apple dropped out. It could be the ballooning valuation, which seems like it might get all the way up to $150 billion. It could be concerns around antitrust. But these are all speculations as neither OpenAI nor

Apple have commented on it. One company that does appear to be set to make an investment is SoftBank. The information reports that they have agreed to invest 500 million as part of the new round. The current crop of potential investors, according to the information, include Thrive Capital in the lead, then Tiger Global Management, Kowatou Management, Microsoft, MGX from the United Arab Emirates, Dragonnear Investment Group, and Nvidia. The information points out that although SoftBank has been a prolific investor in the past, they've been a little bit more circumspect when comes to the most recent wave of generative AI. The investment in Open AI would be the largest bet SoftBank has made in startups working on generative AI. Earlier this year, Masayoshi Sun

vetoed a potential investment in competitor mistral. SoftBank is also invested in perplexity. Now, another one of SoftBank's investments is ByteDance, who are, of course, the owners of TikTok. ByteDance were themselves in the news around AI, planning a new model trained not on NVIDIA chips, but on chips from Huawei. Basically, this is a story around the end results of increasing pressure from the U.S. government to make it harder for Chinese companies to access American chip technology. writes Reuters, BightDance has diversified to domestic suppliers of chips used in AI and accelerated development of its own since the U.S. in 2022 started restricting exports of advanced AI chips such as from market leader, Nvidia. Basically, it sounds like BightDance will be taking

the Ascend 910B chip that they already use for things like inference tasks and trying to make them work for training a new model as well. Yet even without the restrictions on American chip technology, there are still challenges in meeting the demand. Again, according to Reuters, ByteDance has ordered more than 100,000 of these chips, but has so far received fewer than 30,000. On the same topic of Chinese chips maker cambercan technologies was up 20% in trading after Bloomberg reported that the government in Beijing was stepping up pressure on domestic companies to ditch companies like Nvidia and use local alternatives. Basically now, it appears that pressure is coming from both sides, with the U.S. government making it harder to get access to American chips and the Chinese government

pressuring companies to use local alternatives, with the result being that the companies who are in a position to deliver against those shifts are being rewarded in the market. And with that, we will close out today's AI Daily Brief Headlines edition. Next up, the main episode. Today's episode is brought to you by Venice. The leading AI companies store your entire conversation history and attach it to your identity forever. That's every question you ask, every answer you receive, every image you generate, every thought you share with the machine it's all being spied on. If you trust all the company's hackers and NSA board members that will ever have access to your AI conversations, then rejoice, for you are well served. For the rest of us, Venice is an

alternative. Venice is a powerful AI app for text, image, and code generation that respects you as a sovereign individual and believes privacy and free speech are not only human rights, but necessary for civilizational advancement. Private, permissionless, and uncensored, you can try it for free without an account. AIA daily brief listeners receive a 20% discount on Venice Pro. Visit venice.a.com. and enter the discount code, NLW Daily Brief. That's NLW Daily Brief, all one word. Today's episode is brought to you by Super Intelligent, which is, of course, our platform that helps you learn how to use AI tools

and perhaps even more importantly, gives you ideas on the best use cases that are actually going to help you achieve whatever it is you want to achieve. To recognize the end of summer and back to school slash back to work, we are running our best promotion ever. when you sign up for Super Intelligent, using code so back, your first month will be 100% free. The platform features over 600 fun, highly practical AI tutorials that get you using AI fast and with an eye to actually transforming how you get things done.

We've just launched Super for Teams, so if you have a group of people at your company that want to figure out how to use AI together, I highly suggest you check it out. But for those of you who are using Superintelligent as an individual, once again, if you sign up for super intelligent between now and the end of the month using code so back you will get your first month 100% free. Go to be super.a.I. And check it out today. Welcome back to the AI daily brief. After weeks of waiting, California Governor Gavin Newsom has vetoed SB 1047. Of course, for months, SB 1047 has been one of the biggest topics of conversation in AI, certainly the biggest topic of conversation when it comes to AI regulation. The bill would have had a number of

different provisions. It would have required anyone spending more than $100 million to train an AI model to run safety tests and enshrine limits based on the results. It required models to feature a built-in kill switch, allowing human intervention to shut the model down, and it would have potentially created downstream liability for developers of models in a way that opponents of the bill said would have had a freezing effect, especially when it came to open source. There were a few different points that Newsom highlighted in his letter explaining his logic. First of all, he makes it clear why this conversation is so important to California, pointing out that 32 of the world's 50 leading AI companies call California home. He said, as stewards and innovators of the future, I take seriously the

responsible to regulate this industry. Next, he reminded that this was far from the only legislation around AI that had hit his desk. He said this year the legislature sent me several thoughtful proposals to regulate AI companies in response to current rapidly evolving risks, including threats to our democratic process, the spread of misinformation, and deepfakes, risks to online privacy, threats to critical infrastructure, and disruptions in the workforce. On the other hand, he pointed out that SB 1047 was about a different type of risks, specifically about risks that could, but are not guaranteed to emerge. Newsom writes, key to the debate is whether the threshold for regulation should be based on the cost and number of computations needed to develop an AI model,

or whether we should evaluate the system's actual risks regardless of these factors. It seems that a big part of Newsom's rejection is the idea of training and computation as the framework for risk. He claimed that by focusing only on the most expensive and large-scale models, the public could get a false sense of security, when smaller specialized models could emerge that were equally or even more dangerous. He said, while well-intentioned, SB 1047 does not take into account whether an AI system is deployed in high-risk environments, involves critical decision-making, or the use of sensitive data. Instead, the bill applies stringent standards to even the most basic functions so long as a large system deploys it. I do not believe this is the best approach

to protecting the public from real threats posed by the technology. And in here you see some of the core arguments of antagonists of SB 1047 repeated. First of all, an overfocus on hypothetical risks in the future, as opposed to contemporaneous and realized risks right now, as well as a real question around the ultimate arbitrariness of using training costs as the key parameter. Still, what Newsom wasn't arguing was that we should entirely wait for X risks to show up. He said, let me be clear, I agree with the author. We cannot afford to wait for a major catastrophe to occur before taking action to protect the public. Safety protocols must be adopted.

Proactive guardrails should be implemented. And severe consequences for bad actors must be clear and enforceable. Newsom also seemed to suggest that while it may be the case that California has to lead in certain approaches, there's clearly a sense that federal action is on the way and is kind of the more important. thing here. He concludes, I am committed to working with the legislature, federal partners, technology experts, ethicists, and academia to find the appropriate path forward, including legislation and regulation. Given the stakes protecting against actual threats without unnecessarily thwarting the promise of this technology to advance the public good, we must get this right.

Now, interestingly, in addition to that document, the press release that the governor's office sent out, situated this in a wider set of initiatives that he had signed. Indeed, the press release was titled, Governor Newsom announces new initiatives to advance safe and responsible AI protect Californians. The What You Need to Know summary is, Governor Newsom announced that the godmother of AI, Dr. Fay-Fei, as well as a set of others, will help lead California's effort to develop responsible guardrails for the deployment of Gen AI. He also ordered state agencies to expand their assessment of the risks from potential catastrophic events. The press release points out that over the past 30 days, Newsom has signed 17 bills covering deployment and regulation of Gen AI, including those that

would crack down on deepfakes, require AI watermarking, and more. Halfway down the page, after a bunch more information about what he had signed it about this new committee, they added a paragraph quote from the piece we just read about his veto of SB 1047. So what was the response? First, let's look at the side who had been advocating for this. Senator Scott Weiner writes, The veto is a setback for everyone who believes in oversight of massive corporations that are making critical decisions that affect the safety and welfare of the public and the future of the planet. The companies developing advanced AI systems acknowledge that the risks these models present to the

public are real and rapidly increasing. While the large AI labs have made admirable commitments to monitor and mitigate these risks, the truth is that voluntary commitments from industry are not enforceable and rarely work out well for the public. This veto leaves us with the troubling reality that companies aiming to create an extremely powerful technology face no binding restrictions from U.S. policymakers, particularly given Congress's continuing paralysis around regulating the tech industry in any meaningful way. The veto is a missed opportunity for California to once again lead on innovative tech regulation, just as we did around data privacy and net neutrality, and we are all less safe as a result. At the same time, the debate around SB 1047 has dramatically advanced

the issue of AI safety on the international stage. Major AI labs were forced to get specific on the protections they can provide to the public through policy and oversight. Leaders from across civil society from Hollywood to women's groups to youth activists found their voice to advocate for common-sense proactive technology safeguards to protect society from foreseeable risks. The work of this incredible coalition will continue to bear fruit as the international community contemplates the best ways to protect the public from the risks presented by AI. California will continue to lead in that conversation. We are not going anywhere. Dan Hedricks, the director of the Center of AI Safety, and one of the main authors of the bill,

had a similar message. He said that the veto was disappointing and that it represented a reasonable path, but he said, I am not discouraged. The bill encouraged collaboration between industry academics and lawmakers and has begun moving the conversation about AI safety into the mainstream where it belongs. A.I developers are now more aware that they already have to exercise reasonable care lest they be found liable. He talked about how it galvanized a wide coalition of supporters, although he did take a dig at the tech industry saying, discourse and tactics around the bill from some in the industry have been disheartening. It is disgraceful that many opponents trafficked and misinformation to undermine this bill rather

than engaging in a factual debate. SB 1047, he claims, has revealed that some industry calls for responsible AI are nothing more than PR air cover for their business and investment strategies. Now, on the flip side, there was, of course, a ton of excitement. A16Z's Mark Andreessen said, thank you, Governor Newsom for vetoing SB 1047, for siding with California dynamism, economic growth, and freedom to compute over safety, dumerism, and decline. Why, Combinator President Gary Tan, said, what a red-letter day, but truth be told, tomorrow is the beginning of the next fight. The D-cells will never stop. But we won't either. Meta-Chief AI scientist, Jan Lecun, writes, the open source AI community is grateful for your sensible decision, Governor Newsom.

Presidio Bay founder Cyrus said grateful that the rest of the country will see California isn't a lost cause and that common sense can still prevail over misguided ideology. Gavin Newsom made the right call vetoing SB 1047. Scott Weiner would do well to listen to his constituents and support the economic engine driving innovation in his own city rather than pushing for half-baked regulatory measures that would undermine the entire region. What about people who have been a little bit more in the middle? Reed Hoffman writes, appreciate Gavin Newsom for his leadership in recognizing that SB 1047, as written, would have slowed innovation learning and progress towards safe AI by imposing a vague and preemptive regulatory regime on AI developers worldwide, including many

California-based companies. While the bill's authors made a commendable effort to narrow its scope, its final form was still too broad and ambiguous to be workable. To create the models that will make 24-7 access to highly capable and trustworthy medical assistance and tutors, we need regulation that prioritizes responsible innovation, not de facto prohibition through unprecedented liability standards. Mike Noop from Zapier made the important point that even as this debate was going on, we saw one of the challenges of trying to regulate based on today's norms. Mike writes, glad to see SB 1047 officially vetoed. We must keep incentivizing AI innovation not putting speed bumps on last-gen tech. This was under discussed, but frontier models

like OpenAI 01 showed 1047's approach was out of date before it even got decided. Based on his veto letter, Newsom seems to understand this nuance, which should be promising to both frontier researchers and safety advocates. Final thought, I saw many early opponents weaken their opposition as the bill improved. But I saw no early supporters weakened support as it became clear 1047 would not achieve its safety goals. This suggests to me the final bill was largely symbolic, and I think symbolic regulation, a.k.a. bureaucracy has no place at the frontier of progress. Now, at this point, it's very clear that the fight is going to move national. Investor Vindod Kosla writes, great to see smart decision-making by Gavin Newsom on a very important national security and technological

leadership issue. This wasn't a state issue, as Wiener naively assumed. California will continue to lead AI development and hugely help the country. A respondent Sefer, however, wrote, it's not a federal issue either. Safeguards around powerful speech such as AI are a global concern and not one that can realistically be solved. Safety is nigh impossible to define due to a breadth of incompatible perspectives. Mark Andresen went even more aggressive with Vinod, saying Vanad is lobbying to outlaw open source at the federal level. I'm not speculating, I've been in the room as he's doing it. So if you want a preview of whether the next conversation is going to be any less contentious than this one, I don't think so. Now, of course, the next question

question is whether we'll see a veto override or some move to create something like SB 1047 as a California ballot measure. But for now, that seems to be the end of the story for 1047. Certainly, there will be a lot more to come on AI regulation, and of course, I will be here to cover it all. For now, that's going to do it for today's AI Daily Brief. Until next time, peace.