A Draft Republican Exec Order Calls for AI "Manhattan Projects"

Today on the AI Daily Brief, we have a new dueling GOP AI executive order. Before that, in the headlines, a new $100 million fund from Anthropic and Menlo Ventures. 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 kind of a fun one from Menlo and Anthropic. The companies have announced that they are partnering on a new $100 million fund which will support companies that are building on Claude Menlo partner Didi Das writes, Today we're launching the $100 million Anthology Fund, an Anthropic and Menlo Ventures partnership to fund, seed and Series A's as the next generation of AI startups around the world with unique benefits. Startups get $25,000 of anthropic credits, access to anthropic AI models, quarterly deep dives with the Anthropic team, biannual demo days hosted by Anthropic Chief Product Officer Mike as well as Anthropic's co founders, credits from other Menlo Ventures portfolio companies, and more. When it comes to particular thesis areas or startup themes, Didi writes that they're not as much focused on that and instead says that it's, quote, more about attracting high quality, ambitious builders and fueling them with the right resources. Interestingly, they say that you don't have to be using Claude to be in the Anthology fund, but presumably the goal for Anthropic is through these credits and through this access to push people naturally in that direction, which I think is a smart play. I think people will get there. Lastly, this is actually available as an application rather than a traditional pitch process. So if you go to menlovc.com, you can learn more about the fund and access the application. Next up, we move over to the controversial A new report says that, quote, AI models at Apple Salesforce, Anthropic and other major technology players were trained on tens of thousands of YouTube videos without the creator's consent and potentially in violation of YouTube's terms. Basically, the data set in question here is called the Pile. It's a collection by a nonprofit called Eleutherai that was, as described by Ars Technica, put together as a way to offer a useful data set to individuals or companies that don't have the resources to compete with Big Tech. One of the sources for the pile are YouTube captions that are collected by the YouTube's captions API, scraped from over 173,000 videos. Marques Brownlee tweeted, Apple has sourced data for their AI from several companies. One of them scraped tons of data and transcripts from YouTube videos, including mine. Apple technically avoids quote unquote fault here because they're not the ones scraping. But this is going to be an evolving problem for a long time, and the opinions here are complex. On the one hand, there is a sense among many that training without consent is some sort of a violation, even if it's a new type of violation. But others aren't so sure. Darknet Diaries creator Jack Reissetter writes, I optimize for this. I want maximum visibility into my show. We've historically called this SEO, but I wanted to go much further. I want my content to inspire articles, books, newsletters, YouTube videos, podcasts and more. I'm here to make an impact. Micah Berkeley responded to Marques saying, didn't you once say you use GPT when it comes to your content? Which is crazy because the tool that is helping you needs to have knowledge of how the game works. To let me use your expertise for my project, but just don't use my content is wild to me. Deep scanning Google Docs and scraping your website for information is different, but literally reading the transcript of your exceptionally promoted content out in the public I believe makes it fair use. And I think the biggest takeaway from that, even if you disagree with the specifics, is that ultimately this will come down to lines and gray areas. What's more, this is ultimately both a society level question around norms and what people think is acceptable, as well as ultimately a question that I believe will go to the highest levels of the Supreme Court when it comes to how AI and fair use interact. Then again, even as that is being figured out, some people are taking matters into their own hands. The Los Angeles Times reports that a hacker group says it leaked Disney data because of the company's approach to AI. A self proclaimed hacktivist group known as Nullbridge told a set of media outlets, including CNN that it had gotten access to Disney's internal messaging channels and from that leaked roughly a 1.2 terabytes of information. Recently they said that they leaked this internal communication due to the company's handling of quote, artist contracts, its approach to AI and its pretty blatant disregard for the consumer. Nullbulge says that it is a group, quote, protecting artists rights and ensuring fair compensation for their work. So now we have hackers involved in this discussion as well. Recently we talked a little bit about Microsoft's hiring of Inflection AI and their potential mandate around AI agents. And of course we've talked about the fact that their approach to how they got Inflection AI staff to come over was seemingly a way to try to get around antitrust rules. However, that hasn't stopped it from triggering probes, the latest of which is from the UK. The UK's competition and markets Authority has officially opened up an investigation into Microsoft around their hiring of the Inflection AI staff. The CMA now has until September 11th to decide whether or not it would go to the next level of a more in depth investigation. In terms of model launches, Mistral announced a couple new models that have become available, so I expect to hear more about those soon. And lastly, Andrej Karpathy, formerly of both Tesla and OpenAI, has announced a new AI education company called Eureka Labs. This is deserving of a much larger episode and we will probably do a full episode on it at some point, but for now let's just read Andre's announcement piece. He writes, we at Eureka Labs are building a new kind of school that is AI native. How can we approach an ideal experience for learning something new? For example, in the case of physics, one could imagine working through very high quality course materials together with Feynman who is there to guide you every step of the way. Unfortunately, subject matter experts who are deeply passionate, great at teaching, infinitely patient and fluent in all the world's languages are also very scarce and cannot personally tutor all 8 billion of us on demand. However, with recent progress in generative AI, this learning experience feels tractable. The teacher still designs the course materials, but they are supported, leveraged and scaled with an AI teaching assistant who is optimized to help guide the students through them. This teacher plus AI symbiosis could run an entire curriculum of courses on a common platform. If we are successful, it will be easy for anyone to learn anything, expanding education in both reach and extent. Andre goes on to say that their first course will be LLM101, an undergraduate level course that guides the student through training their own AI. And perhaps most interestingly, he says, Eureka Labs is the culmination of my passion in both AI and education for over two decades. He said, all of my work combining the two so far has only been part time as side quests to my quote unquote real job. So I'm quite excited to dive in and build something great professionally and full time. It is pretty awesome to see someone of Andre's caliber bringing the full force of his attention to education and I think it will be very exciting to see what they build. For now though, that is going to do it for the headlines. Up next, the Main Episode Today's episode is brought to you by superintelligent, the platform for fun fast AI Learning super has a ton of new things going on. We recently announced our partnership with Spotify, through which users of that app can now access super intelligent content directly from their mobile apps. We've also just launched the AI Learning Feed. In addition to seeing the tutorials that we're dropping, there are polls, news items with related lessons, and a chance for people to show off the projects and use cases that are making AI come alive for them. We've also just kicked off the Super Summer Challenge where each week we'll share a new challenge that you can use to discover new AI tools and use cases. Go to BeSuper AI and use code SUPERFUND for 50% off your first two months. That's BeSuper AI. 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. 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 companies, 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 co generation that respects you as a sovereign individual and believes privacy and free speech are not only human rights, but are necessary for civilizational advancement. Private, permissionless and uncensored, you can try it for free without an account at Venice AI. Welcome back to the AI Daily Brief. Quick caveat which I will keep giving for a little while. When it comes to political related episodes, times are of course incredibly tense right now when it comes to American presidential politics. The AI Daily Brief is of course not a political show. It is a show about artificial intelligence. However, the more that AI becomes a political issue, the more that we will have to cover political issues. My promise is to focus strictly on the AI dimensions of those conversations rather than the broader political discourse. With that out of the way, just a day after nominating J.D. vance as his Vice president, it appears that the Trump camp has prepared a draft executive order on artificial intelligence. Today we're going to talk about what we know so far about that, as well as the context and where it came from where our story begins is last October when we got President Biden's executive order on artificial intelligence. There were a few things that stood out very clearly to people. Perhaps the biggest was the Biden administration using the Defense Production act as justification to require that developers of the most powerful AI systems share their safety test results and other critical information with the US Government. There was more that pushed people in this direction, but one of the critiques coming out of that was that a lot of the compliance requirements here were likely to preference and prioritize the incumbents over new challengers. AI entrepreneur Jeff Amico said Biden's AI executive order is out and it's terrible for US innovation. Here are some of the new obligations which only large incumbents will be able to comply with. He continues, if you're developing a large model, you must report to the government regarding any ongoing or planned activities related to training, development or producing your model. Essentially public company reporting for startups building large models. He continues, if you are developing a large model, you must report the quote ownership and possession of the model weights of your model and the physical and cybersecurity measures taken to protect those model weights, end quote, which Jeff says is an implied restriction on open sourcing model weights. Another thing that many people had problems with was the provision that quote, companies, individuals and other organizations that acquire developer possess a potential large scale computing cluster have to report any such acquisition, development or possession, including the existence and location of these clusters and the amount of total computing power available in each cluster. Jeff argues that this is treating compute an inherently neutral technology as a dangerous resource that must be regulated. Now, the takes were of course varied about this. There were many who felt like Jeff did this was bad for innovation. There were others who thought that it was fairly middle of the road and argued that while it's not perfect, having insight and visibility into where people were accessing large amounts of compute might be the least burdensome way for the US Government to maintain visibility into the frontiers of AI development. Arguably the next big piece of AI policy news came in May of this year when a group of bipartisan senators led by Chuck Schumer released its AI roadmap. This one was a disappointment to many Democrats. The document, as I've discussed previously on this show, clearly prioritized AI leadership over what some would have seen as AI safety consideration, leading to takes like this one from economics professor Hal Singer, who wrote Schumer offers a quote unquote roadmap for quote unquote regulating AI that calls for 32 billion in funding AI development and individual sectors to develop their own rules for AI effectively self regulation. Bravo. This is the opposite of governing. Economist and columnist Tyler Cowan responded to Schumer's piece with an op ed in Bloomberg called the AI Safety Movement is Dead. He argued that this represented a turning point. He wrote, Senate Majority Leader Chuck Schumer and his working group on AI have issued a guidance document for federal policy. The plans involve a lot of federal support for the research and development of AI and a consistent recognition of the national security importance of the US Maintaining its lead in AI. Lawmakers seem to understand that they would rather face the risks of US based AI systems than to have to contend with Chinese developments without a US Counterweight. That was very much my takeaway as well. Then just about a week and a half ago, we got the official GOP platform. It included a short section on innovation policy that included discussions of crypto space and AI. The AI section read, we will repeal Joe Biden's dangerous executive order that hinders AI innovation and imposes radical left wing ideas on the development of this technology. In its place, Republicans support AI development rooted in free speech and human flourishing. Now of course, that on its own doesn't mean much other than they didn't like what they saw in the executive order from Biden. But now with JD Vance announced as VP candidate who has been loudly outspoken in favor of open source AI and against incumbent regulatory capture, we're now starting to get hints of what AI policy under a new Trump administration might look like. The Washington Post yesterday published Trump Allies Draft AI Order to Launch Manhattan Projects for Defense. The plan to, quote, make America first in AI and roll back, quote, burdensome regulations would favor Silicon Valley investors who are now flocking to support the former president. A couple things to note. First of all, we have not seen this executive order. It is still apparently in a draft stage, but that draft was viewed by the Washington Post. Some of the additional details from the Post they say that the EO would create, quote, industry led agencies to quote, evaluate AI models and secure systems from foreign adversaries. And the EO suggest that they would also launch a series of what they call Manhattan projects to develop military technology and immediately review what they call unnecessary and burdensome regulations. That is mostly the substance that we got from the Post piece on this eo. When asked for comment, the Trump campaign pointed to a blog post which said that no aspect of future presidential staffing or policy announcements should be deemed official unless they come directly from Trump or his campaign team. And a spokesperson from the America First Policy Institute, said in a statement that the document does not represent the organization's, quote, official position. The big connection that the Post makes was between these new AI plans being leaked and the amount of involvement that the tech industry seems to increasingly be playing in the Trump campaign. They point to, for example, a 90 minute podcast that was just released as a discussion between Marc Andreessen and his partner Ben Horowitz. The podcast lays out why they've decided to wade more aggressively into politics, what the substance of their, quote, little tech agenda is, their articulation of the importance of AI in America and why they're supporting Trump over Biden. Twitter user Joyce park summed up a lot of the skepticism out there when she tweeted, wow, who would have predicted that the day after Musk, Andreessen and Thiels Boy all pledged to Trump, suddenly AI deregulation is on the top of his agenda? Professor Ethan Moloch, meanwhile, pointed out, quote, for what it's worth, meta has already spent almost as much as the Manhattan Project on GPUs, inflation adjusted. And from that statistic alone, you know that this is not an issue that is likely to go away. For now, though, that's all we have to go on. So we will have to wrap it there. I will of course, watch very closely to see if we get any more information around what AI policy might look like either under a new Trump administration or a second Biden administration. For now, that's where we will wrap. Appreciate you listening or watching as always. And until next time, peace. Sat.