Brian McCullough (0:04)

Welcome to the TechMe write home for Wednesday, February 12th, 2025. I'm Brian McCullough. Today the first big AI legal ruling has come down, and it might not be good news for AI startups. Apple launches its biggest health study yet this one weird trick that tech companies are using to make their AI spending seem not so expensive. And Matt Levine on the whole Elon buying OpenAI thing. Here's what you missed today in the world of tech Foreign Thomson Reuters has won the first major USAI copyright ruling against fair use in a case filed in May 2020 against legal research AI startup Ross Intelligence. Quoting Wired In 2020, the media and technology conglomerate filed an unprecedented AI copyright lawsuit against the legal AI startup Ross Intelligence. In the complaint, Thompson Reuters claimed the AI firm reproduced materials from its legal research firm, Westlaw. Today, a judge ruled in Thomson Reuters favor, finding that the company's copyright was indeed infringed by Ross Intelligence's actions. None of Ross's possible defenses hold water. I reject them all, wrote U.S. district Court of Delaware Judge Stefanos Bibas in a summary judgment. The fair use doctrine is a key component of how AI companies are seeking to defend themselves against claims that they use copyrighted materials legally. The idea underpinning fair use is that sometimes it's legally permissible to use copyrighted works without permission, for example to create parody works or in non commercial research or news production. When determining whether fair use applies, courts use a four factor test looking at the reason behind the work, the nature of the work, whether it's poetry, Nonfiction, private letters, etc. The amount of copyrighted work used, and how the use impacts the market value of the original. Thomson Reuters prevailed on two of the four factors, but Beavis described the fourth as the most important and ruled that Ross quote meant to compete with Westlaw by developing a market substitute. End quote. Now worth noting that this is an AI tool that is not generative. This was not an LLM. But James Gimmelman, an Internet law professor, told Wired, if this decision is followed elsewhere, it's really bad for the generative AI companies. Quoting the Verge, Thomson Reuters sued over Ross use of its Westlaw search engine. Westlaw indexes a good deal of material that is not copyrightable, like legal decisions, but also intersperses it with its own content. For instance, Westlaw headnotes, which are summaries of points of law written by human editors, are a signature feature meant to make the very expensive Westlaw subscription attractive to lawyers. In building a legal research search engine, Ross turned the annotations and headnotes into numerical data about the relationships among legal words to feed its AI, wrote bbas. Neuraling describes how after Thompson writers rejected its attempt to license Westlaw's content, Ross turned to another company, legalese, and purchased 25,000 bulk memos of questions and answers written by lawyers using the Westlaw headnotes that it used for training data. End quote and quoting Sherwood News the original complaint claimed that Thomson Reuters Westlaw legal research database was, quote, illicitly and surreptitiously used to acquire access to and copy plaintiffs valuable content to create Ross Intelligence's AI powered tool in an alleged violation of the publisher's copyright. End quote so not a lawyer, but it sounds like the key here is the acquisition of the data, the hoovering up of the data not being fair use, which would sound like it would apply to how LLMs get their data. Also, Apple has partnered with Brigham and Women's Hospital to launch the voluntary Apple Health Study via its research app. Open to all iPhone users in the US Quoting cnet this extensive, virtual and completely optional study will collect and analyze Apple users health metrics such as sleep, heart and respiration rates, activity levels and more. Apple aims to further understand the connection between technology and social, mental and physical health, and whether subtle personal health changes can signal something larger. Apple has already conducted three studies through its research app, the Apple Heart and Movement Study, the Apple Women's Health Study and the Apple Hearing Study. These studies had a combined total of about 420,000 participants. The data collected paved the way for new iPhone health features. The Apple Health Study won't focus solely on one area of health. Instead, it'll take a holistic view of wellness, including but not limited to mental health, activity level, cardiovascular health, metabolic health, mobility, neurologic health, respiratory health, sleep, and more. In addition to data, the study will also ask participants questions about lifestyle and habits. This will allow researchers to explore the connections between all areas of wellbeing such as sleep and exercise, mood and heart rate, and so on. Participation allows Apple to evolve over time because insights gathered may influence future product development and Quoting the Verge the virtual study will appear in the research app and is being conducted in conjunction with Brigham and Women's Hospital, a Harvard Medical School affiliate. Apple also partnered with the hospital for its heart and movement study. Users who opt in to the study will not only contribute their data, but also answer periodic survey questions about their at home life and habits. The study will encompass topics including activity, aging, cardiovascular health, circulatory health, cognition, hearing, menstrual health, metabolic health, mobility, neurological health, respiratory health and sleep to start. The study is set to last about five years and could potentially be extended further. End quote and quoting CNBC Apple has no access to participants identifiable information, the company said. We've only just begun to scratch the surface of how technology can improve our understanding of Human Health, Dr. Callum McRae, the principal investigator of the study at Brigham and Women's Hospital, said in a statement. End quote Apple Research is available in the App Store for the iPhone and Apple Watch. The Apple Health study was added to the app this morning. Sources are telling Forbes that Meta is in talks to acquire South Korean AI chip startup Furiosa AI, which has raised around $115 million since its 2017 founding. The deal could close this month. Apparently Meta is not the only bidder to acquire this company. Quot the Seoul based startup last raised funds just last week when it received 2 billion won, about $1.4 million from CRIT Ventures, a South Korean venture capital firm founded by Jaejoong Song, former CEO of online gaming company com2us. A valuation for the company was not disclosed. In total, Furiosa AI has raised about 170 billion won, or about $115 million in venture funding. Its early investors include South Korean Internet giant Naver and sole based DSC investment. Furiosa AI was founded in 2017 and is led by Jun Paik, who previously worked at Samsung Electronics and amd. As of the most recent public filing, pike owned an 18.4% stake in Furiosa AI. In August, Furiosa AI unveiled its RNGD chip, which was developed in partnership with Taiwanese custom chip maker Global Unichip Corp. While cutting edge GPUs consume up to 1200 watts, RNGD pronounced Renegade is designed to operate with a thermal design power or TDP of just 150 watts, Furiosa AI said in a blog post about Renegade. This makes Renegade an ideal choice for large scale deployment of advanced generative AI models like Llama 2 and Llama 3. Llama 2 and Llama 3 are Meta's publicly available large language models. You want to know what the bottom line is here? Furiosa AI says that its Renegade chip has 3x better performance per watt than Nvidia's H1 hundreds. There you go, cost savings and the Renegade chip is slated to go into mass production later this year. Everybody wants their own custom chips. Expect this sort of deal to be the beginning of a land grab. Ever feel overwhelmed trying to manage your web hosting while juggling a million other tasks? Even if tech isn't your thing Kinsta's managed WordPress hosting is a relief. Their expert team handles it all. They've bundled up all the essentials to make sites stress free with speeds that'll wow your visit, security that never sleeps, and a dashboard so intuitive you'll wonder why everything isn't this easy. Trust me, I know how confusing outdated hosting dashboards are. Kinsta's custom control panel is intuitive and doesn't feel like it's from the last decade. Kinsta doesn't just host WordPress websites, they deliver blazing speed, ironclad security and reliability. Worried about moving? They'll migrate your entire website for free and you've got a 30 day money back guarantee. Imagine your site running at warp speed. Just by switching to Kinsta, your website could run up to 200% faster. TripAdvisor, NASA and Indeed are among the 120,000 businesses that trust Kinsta with their WordPress websites. Tired of being your own website support team? Switch your hosting to Kinsta and get your first month free. And don't worry about the move, they'll handle the whole transition for you. No tech expertise required, just visit kinsta.com Brian to get started. That's K-I-N-S-T A.com B R I A N Ready to optimize your nutrition this year, Factor has chef made gourmet meals that make eating well easy. They're dietitian approved and ready to heat and eat in two minutes so you can fuel right and feel great no matter what life throws at you. I told you this is my wife's daily lunch solution. She's got these factors in the office and boom. No leaving work to find a bite to eat. Factor arrives fresh and fully prepared, perfect for any active, busy lifestyle. Lose up to 8lbs in 8 weeks with factor Keto Meals Based on a randomized controlled clinical trial with Factor Keto. Results will vary depending on diet and exercise. With 40 options across eight dietary preferences on the menu each week, it's easy to pick meals tailored to your goals. Choose from preferences like Calorie Smart, Protein plus or Keto. And these are tasty. Believe me, I have them too. Just don't tell my wife. Factor can help you feel your best all day long with wholesome smoothies, breakfasts, grab and go snacks and more add ons. Reach your goals this year with ingredients you can trust and convenience that can't be beat. Eat smart with Factor get started@factor meals.com Ride50OFF and use code Ride50OFF to get 50% off your first box plus free shipping. That's code RIDE50OFF@factormeals.com Ride50OFF to get 50% OFF plus free shipping on your first box. This is a cute little trick. In Q4, Meta tweaked its depreciation accounting for AI infrastructure, extending it from the previous four to five years to now five and a half years, thereby cutting expenses by $2.9 billion in 2025 alone and boosting profit. Quoting Bloomberg, it wasn't the release of a new product or cost cuts. It was a tweak to an accounting formula used to measure the depreciation of its expensive artificial intelligence infrastructure. The change, disclosed in the social media giant's earnings materials on January 29th extends the so called useful life period of certain servers and networking assets to five and a half years from the four to five years it previously used. While that may sound trivial, its impact on earnings will be sizable given the heavy spending on these relatively short lived assets. By Meta's reckoning, the shift is expected to reduce the company's depreciation expense by $2.9 billion in 2025, which would on its own amount to almost 4% of the estimated pre tax profits for the year. With Meta planning to spend as much as 75% more this year on capital expenditures to build out its AI capabilities, the effect will be even bigger in 2026. The changing expectations highlight how companies are grappling with the temporary shelf life of the tens of billions of dollars of new semiconductors and computer servers they are to power their AI services. Meta is now hoping that the equipment will last longer than they had expected. While there may be legitimate reasons to extend the server life based on their actual experience, it also decreases the depreciation in the short run and improves the bottom line, said Ravi Gottiman, partner at tax and accounting firm Ziyon Research Group. Meta's chief financial officer Susan Lee, said on the most recent earnings call that the company is making efficiency gains. Quote by extending the useful lives of our servers and associated networking equipment. A Meta representative declined to comment. Meta isn't alone in changing its timetable on depreciation and with it, the financial results. In 2022, Microsoft extended the useful lives of server and networking equipment to six years from four. In 2023, Oracle extended its estimate to five years from four, according to a filing. Others, however, have taken the opposite approach. Amazon said this month that the lifespan of the equipment is growing shorter from six years to five. The change, which took effect on January 1, will cut operating income by about $700 million. The company said in a filing end quote on that whole Elon OpenAI thing from yesterday, sources say Sam Altman has told staff that OpenAI's board has not actually received a formal offer from Elon Musk and other investors yet, but if it happens, the board plans to reject it. Quoting Wired internally, OpenAI employees reacted to the news with a mixture of fear and exasperation. While the nonprofit board doesn't have a fiduciary responsibility to maximize returns for investors, it does have a duty to negotiate a reasonable valuation of OpenAI's assets to pursue the company's nonprofit goals. If the board took a lower offer from Altman or a company he controls, it would likely be breaching its fiduciary duty since Altman is considered an insider, says Samuel D. Brunson, a law professor at Loyola University Chicago who spent specializes in nonprofit organizations. OpenAI did not respond to a request for comment from Wired. Elon's bid establishes a floor for the value of those assets, Brunson says. At the very least, it makes it much more complicated for OpenAI to spin off the assets into a for profit controlled by Sam Altman, end quote. That's an interesting wrinkle there. And also this is the crux of the matter, right as we talked about yesterday. But if you've been a listener long enough, then you'll remember that when Elon was buying Twitter, one of my go tos for figuring that whole mess out was Matt Levine. Well, guess who's back and on fire again. Quoting from Matt's most recent newsletter. So you are left with the absolutely bizarre circumstance that a nonprofit plausibly might have a fiduciary obligation to sell to the highest bidder, even if it finds that highest bidder unsavory and uncharitable. What if it got a topping bid from the Chinese Communist Party? What if a robot wearing a fake mustache came in and said I will pay $150 billion for your company and will not use it to take over the world and enslave humanity. But even gave you that idea with the charity's obligation, its obligation not to give assets away to a for profit company, but to be paid fair value for them, require it to sell to the highest bidder. And recognizing that bizarre circumstance, Musk, the richest man in the world, lobbed in a bid. I cannot fault it. It is top tier M and a trolling. One assumes he is not serious. I should say everyone, including me and Sam Altman, assumes that Musk will not actually buy OpenAI, but he would do it for $97.4 billion. That's an incredible bargain. If he bought the nonprofit's controlling stake for $97.4 billion, he'd control the business. OpenAI of the nonprofit wants to give up control of the business because that will help with fundraising. But Musk wouldn't. Musk would pay $97.4 billion to get control of a 260 billion-ish business. He'd have minority shareholders or weird profit interest holders, but he could deal with them and he'd have no trouble with raising. The point here is not to buy OpenAI. The point is to raise the price. If the right valuation of the nonprofit stake in OpenAI is north of $100 billion, then that means that the nonprofit needs to get a bigger chunk of the company, which leaves less for the for profit shareholders. It is harder to raise $40 billion from SoftBank if that $40 billion gets SoftBank a smaller share of the company. The nonprofit board's obligation is to its mission and it should make decisions that are in the best interest of building AI for the benefit of humanity, not to maximize dollars. On the other hand, OpenAI kind of literally is for sale in that the nonprofit really is not allowed to give it away to a for profit company. Someone has to buy OpenAI the business from OpenAI, the nonprofit. OpenAI would like that someone to be OpenAI the business. But OpenAI has put itself in play and now there's a higher bidder. And again, as with Twitter, the problem here, the thing that leaves the board vulnerable to Musk, is the fuzziness of OpenAI's vision. If OpenAI had a vision like we are going to build OpenAI for the benefit of mankind rather than for Prof. Then there would be absolutely no question of it selling out to the highest bidder. That would be an absurd suggestion. Or if OpenAI had a vision like, we are going to make zillions of dollars building AI, it's going to be absolute skynet up in here. There would be no question of it selling to Elon Musk for any price even he could name. But OpenAI is trying to do both. It is still pursuing its benefit of humanity mission while also converting it to a for profit that is messy and arguably insupportable. And Musk pounced, by the way. I assume he does not have $97 billion of committed financing. Like, number one. That would be a lot. By a factor of two, the largest ever leveraged buyout, and by a factor of, oh, probably infinity, the largest ever leveraged buyout of a nonprofit. Number two, honestly, how would you even get financing to do a leverage buyout of a nonprofit? I'm sure there's an answer, but. But man number three, this is mostly trolling. Like, the goal here is not to buy OpenAI, but to make it awkward and difficult for OpenAI to convert to a for profit company. Seems silly to get commitment letters for this. At the same time, though, he's not kidding, right? Like, he clearly could raise $97 billion for a stunt like this. I suppose OpenAI could reject his offer for the uncertainty of his financing, and then he'd come back 12 hours later and be like, aha, I have the money right here. And then where would they be? He'll roll up with 9.7 trillion pennies, won't he? End quote. Nothing more for you today. Talk to you tomorrow.