Tech Brew Ride Home - Mon. 11/11 – FTX Sues Everybody

Episode Date: November 11, 2024

FTX is suing basically anybody you can think of. Why will your iPhone now reboot itself if you don’t log into it for several days? Is OpenAI’s next flagship model underperforming what they were expecting? And I’ve found the one company most disrupted by AI. At least, so far. Sponsors: 1Password.com/ride Links: FTX Sues Binance, Ex-CEO Zhao Seeking $1.8 Billion Clawback (Bloomberg) FTX Sues Scaramucci to Recoup Money for Creditors (Bloomberg) Apple Quietly Introduced iPhone Reboot Code Which is Locking Out Cops (404Media) OpenAI Shifts Strategy as Rate of ‘GPT’ AI Improvements Slows (The Information) How ChatGPT Brought Down an Online Education Giant (WSJ) Painting by A.I.-Powered Robot Sells for $1.1 Million (NYTimes) Learn more about your ad choices. Visit megaphone.fm/adchoices

On April 4th, 2023, around 2 in the morning, a man was found stabbed multiple times on a sidewalk in downtown San Francisco. Hey, who did this to you? What happened next turned the story into a political firestorm. Reports have identified the victim as Bob Lee, the founder of Cash App. From Bloomberg Podcasts, this is Foundering, the Killing of Bob Lee, beginning April 16. Welcome to the Tech meme right home for Monday, November 11th, 2024. I'm Brian McCullough today. FtX is suing basically anybody you can think of. Why will your iPhone now reboot itself if you don't log into it for several days?

Is Open AI's next flagship model underperforming what they were expecting? And I've found the one company most disrupted by AI, at least so far. Here's what you missed today in the world of tech. FTX is suing, well, everybody? Let me run down this list. is suing Binance, the company itself, but also former CEO CZ, seeking to claw back almost $1.8 billion, it alleges, was fraudulently transferred by Sam Bankman-Fried in a 2021 deal, quoting Bloomberg. Binance Zhao and other Binance executives received the funds as part of a July 2021 share purchase

deal with Bankman Fried, the FDX co-founder who is now in prison. In that transaction, they sold stakes of about 20% in FTX's international unit and 18.4%. in its U.S.-based entity, according to a legal filing from the FTX estate on Sunday. Bankman Freed paid for the stock repurchase using a mix of FTX's exchange token FTT and Binance-branded coins, BNB and BUSD, valued at $1.76 billion at the time, according to the filing. FDX and its sister trading house Alameda Research, quote, may have been insolvent from inception and certainly were balance sheet insolvent by early 2021, the estate said in the filing. As a result, repurchase deal was made fraudulently, it alleged. FDX also accused Zhao of posting a series of, quote,

false, misleading and fraudulent tweets shortly before FTCS's collapse, the content of which was, quote, maliciously calculated to destroy his rival. A November 6, 2020 tweet by Zhao stated that Binance intended to sell its FTT tokens worth some $529 million at the time, causing withdrawals from the exchange to skyrocket. The claims are meritless, and we will vigorously defend ourselves a Binance spokesperson said in a statement on Monday. A representative for Zhao didn't immediately reply to an emailed request for comment, end quote. Next up, FtX is suing Anthony Scaramucci, crypto.com, Mark Zuckerberg's forward.us, and others to recoup SBF's investments during his campaign of influence buying in 2022, according to the suit, also Bloomberg.

quote, FTCS alleges that during the crypto winner of 2022, founder Sam Bankman-Freet engaged in, quote, a campaign of influence buying throughout the year and making lavish and showy investments. One connection that Bankman Fried poured significant time and money into was Scaramucci for his, established financial, political, and social network, according to the filing. FTX is now going after these investments as it claims they, quote, conveyed little to no benefit and, quote, instead served only to prop up Bankman Freed's standing in the world of politics, and traditional finance. The bankrupt crypto firm alleges that Bankman Freed invested $67 million into various Skybridge capital endeavors, the firm run by Scaramucci in 2022, as Scaramucci had been,

quote, seeking a bailout. Skybridge's assets under management had fallen from a 2015 high of $9 billion to $2.2 billion, according to the filing. FDX is now seeking to recover more than $100 million in damages, end quote. Finally, for good measure, they are suing the crypto trader known as Humpy the whale, alleging margin trading, market manipulation from January 2021 to September 2022, leading to FTX and Alameda losing $1 billion. Folks have been noticing that if you don't unlock your iPhone for a period of days, thanks to a new feature in iOS18.1, it will completely reboot itself. Why is this a feature?

Well, it's kind of interesting. Quoting 404 Media. Apple quietly introduced code into iOS18.1, which reboots the device if it has not been unlocked, for a period of time reverting it to a state which improves the security of iPhones overall and is making it harder for police to break into the devices, according to multiple iPhone security experts. On Thursday, 404 media reported that law enforcement officials were freaking out that iPhones, which had been stored for examination, were mysteriously rebooting themselves. At the time, the cause was unclear, with the officials only able to speculate why they were

being locked out of the devices. Now a day later, the potential reason why is coming into view. quote, Apple indeed added a feature called Inactivity reboot in iOS18.1, Dr. Ing Jiska Klasen, a research group leader at the Hasso Platner Institute, tweeted after 404 media published on Thursday, along with screenshots that they presented as the relevant pieces of code. In a law enforcement and forensic expert-only group chat, Christopher Vance, a forensic specialist at Magnet Forensics, said, we have identified code within iOS 18 and higher that is an inactivity timer. This timer will cause devices in an AFU state to reboot to a BFU state after a set period of time, which we have also identified, end quote. AFU refers to after first unlock, which is when somebody, presumably the phone's owner, has unlocked the device at least once since being powered on,

and which generally can make it easier for law enforcement to unlock. BFU or before first unlock is when a user has not unlocked the phone since it was turned on and is typically a harder state for forensic tools to crack. The reboot timer is not tied to any network or charging functions and only tied to inactivity of the device since last lock. The researchers wrote, 404 Media obtained multiple screenshots of Vance's messages in the group chat from a source. 404 Media granted them anonymity because members are typically not allowed to share communications from this group chat. Magnet Forensics recently acquired Gray Shift, the company that makes the phone unlocking tool Gray Key, Rick Andrade, a spokesperson for Magnet Forensics, declined to comment.

We can't comment on specific issues, but as Chris said, we're looking into it, he wrote in an email. Chris Wade, the founder of Mobile Analysis Company Correllium, told 404 Media that after the fourth day of a device being in a locked state, the device reboots. Apple did not respond to multiple requests for comment about the reboots and the inactivity feature sent on Thursday and Friday. The iOS change is the latest skirmish in the ongoing battle between phone manufacturers like Apple, whose main motivation is protecting their users' data, and forensic firms and law enforcement who want to extract data from seized devices. Initially, the law enforcement officials raising the alarm about the rebooting iPhones speculated that the lockouts were due to their seized iPhones not being connected to a cellular network, or perhaps even an iOS 18 device somehow telling

other nearby iPhones to reboot themselves. The real explanation, based on what the multiple experts found appears to be more about a certain amount of time passing rather than anything else. Remember that the real threat here is not police. It's the kind of people who will steal your iPhone for malign purposes. Matthew Green, a cryptographer and associate professor at Johns Hopkins University, told 404 Media, this feature means that if your phone gets stolen, the thieves can't nurse it along for months until they develop the tech to crack it, end quote. Green called the feature a huge improvement in terms of security, he added, I would bet that rebooting after a reasonable inactivity period probably doesn't inconvenience anyone, but does make your phone

a lot more secure. So it seems like a pretty good idea, end quote. The information says that Orion, Open AI's upcoming flagship model currently in development, is showing a smaller increase in quality from GPT4 than GPT4 had over GPT3 at launch. Quote, the number of people using chat GPT and other artificial intelligence products is soaring. The rate of improvement for the basic building blocks underpinning them appears to be slowing down, though. The situation has prompted OpenAI, which makes ChatGPT to cook up new techniques for boosting those building blocks known as large language models to make up for the slowdown. The challenges OpenAI is experiencing with its upcoming flagship model codenamed Orion show what the company is up against. In May, OpenAI CEO Sam Altman

told staff he expected Orion, which the startup's researchers were training, would likely be significantly better than the last flagship model released a year earlier. Though OpenAI had only completed 20% of the training process for Orion. It was already on par with GPT4 in terms of intelligence and abilities to fulfill tasks and answer questions, Altman said, according to a person who heard the comment. While Orion's performance ended up exceeding that of prior models, the increase in quality was far smaller compared with the jump between GPT3 and GPT4, the last two flagship models of the company released, according to some OpenAI employees who have used or tested Orion. Some researchers at the company believe Orion isn't reliably better than its predecessor in handling

certain tasks according to the employees. Orion performs better at language tasks, but may not outperform previous models at tasks such as coding, according to an OpenAI employee. That could be a problem, as Orion may be more expensive for OpenAI to run in its data centers compared to other models it has recently released, one of the people said. The Orion situation could test a core assumption of the AI field known as scaling laws, that LLMs could continue to improve at the same pace as long as they had more data to learn from, and additional computing power to facilitate that training process. In response to the recent challenge to training-based scaling laws posed by slowing GBT improvements, the industry appears to be shifting its efforts to improving models after their initial

training, potentially yielding a different type of scaling law. Some CEOs, including meta-platforms Mark Zuckerberg, have said that in a worst-case scenario, there would still be a lot of room to build consumer and enterprise products on top of the current technology, even if it doesn't improve. At OpenAI, for instance, the company is busy baking more codewriting capabilities into its models to head off a major threat from rival Anthropic, and its developing software that can take over a person's computer to complete white-collar tasks involving web browser activities or applications by performing clicks, cursor movements, text, typing, and other actions humans perform as they work with different apps. Those products, part of a movement toward AI agents that handle multi-step tasks, could prove just as revolutionary as the initial launch of ChatGPT. Furthermore, Zuckerberg Altman and CEOs of other AI developers also publicly say they haven't hit the limits of traditional scaling laws yet, end quote. All right, I think I've found it, the one company that has seemingly been the most disrupted by this current AI generation. The company Chegg has lost more than 500,000 subscribers since ChatGPT's launch, and its stock is down 99% from early 2021, as students looking for homework help turned to free A.

tools instead of tools like Chegg, quoting the journal. The online education company was for many years the go-to source for students who wanted help with their homework or a potential tool for plagiarism. The shift to virtual learning during the pandemic sent subscriptions and its stock price to record highs. Then came chat GPT. Suddenly students had a free alternative to the answers Chegg spent years developing with thousands of contractors in India. Instead of Chegging, the solution, as it was known, they began canceling their subscriptions and plugging questions into chatbots.

Since ChatGPT's launch, Cheg has lost more than half a million subscribers who pay up to 1995 a month for pre-written answers to textbook questions and on-demand help from experts. Its stock is down 99% from early 2021 at racing some $14.5 billion of market value. Bond traders have doubts the company

will continue bringing in enough cash to pay its debts. Though Chegg has built its own AI products, the company is struggling to convince customers and investors, it still has value in a market upended by ChatGPT. It's free, it's instant, and you don't really have to worry if the problem is there or not, Jonah Tang, an NBA candidate at Point Loma Nazarene University in San Diego, said of the advantages of using ChatGPT for homework help over Chegg. A survey of college students by Investment Bank Needham found 30% intended to use Chegg this semester,

down from 38% in the spring, and 62% plan to use ChatGPT, up from 43%. My concern is that the headwinds to Chegg's top line aren't temporary. They're more structural in nature, said Neenam analyst Ryan McDonald. Dan Rosenzweig, Chegg's CEO of more than a decade, stepped down in June after the stock cratered under his leadership. Chegg said Rosenzweig notified the board of directors that he planned to retire a year in advance. Researchers at the University of Illinois at Urbana-Champaign conducted a study in the spring of last year to see how ChatGPD had influenced cheating in an introductory programming course. They found students had overwhelmingly moved to chat GBT from what the researchers called plagiarism hubs such as Chegg. It appeared that they completely shifted over from trying to find online solutions and copying them to just going to chat GPT and having it generate solutions for them, said Craig Ziles, Professor of Computer Science at the University of Illinois at Urbana-Champaign. Chegg began as a message board for Iowa State University students with a name that combined chicken and egg.

It evolved into a textbook rental company during the 2000s and began offering human-created online study guides in the 2010s, end quote. Finally, today a painting depicting Alan Turing as the god of AI, which was created by an AI-powered humanoid robot called Ada, sold at Sotheby's auction for nearly $1.1 million, quoting the Times. The experiment was the brainchild of Aidan Meller, a former gallerist living outside Oxford, England, who has worked with a team of, of nearly 30 people to build the robot. In most recent appearances, the robot is dressed like a woman with a bob haircut and is referred to as Ada in honor of Ada Lovelace, the 19th century mathematician who has been recognized as the world's first computer programmer. I am trying to adapt to this slightly surreal moment, Miller said in an interview recalling the final moments of the sale. The painting, which depicted Turing as the god of artificial intelligence, was offered as part

of Sotheby's digital art sale and initially was estimated to sell for 120, to $180,000. It received more than 27 bids and was sold to an anonymous buyer from the United States. Meller said the proceeds from the sale of the painting called AI God Portrait of Alan Turing would help finance new improvements to Ada's design. We plow everything back into the project, he said. She is constantly being updated. She is on her third painting arm already. Mellar originally prompted Ada to create something for a conference on artificial intelligence organized by the United Nations this year. The robot responded with a suggestion that it paint a portrait of Turing as an example of someone

who predicted the power of AI technology as early as the 1950s. But the process of actually finishing the work was more complicated. Ada's programming interpreted a photograph of Turing and produced 15 individual paintings based on different parts of his face. The robot chose three of the portraits alongside a painting it had made of the bomb machine, the large device that Turing and other codebreakers used to decrypt ciphers, generated by Nazi Germany's Enigma machine. The works were then photographed and uploaded to a computer that used Ada's language model to decide on the assembly of a single painting, which was then

completed using a 3D textured printer. Studio assistants helped to create a more realistic finished product on the canvas. Ada then added marks and textures to the portrait to complete it. Miller said Ada was supposed to prompt discussions about the ethics of artificial intelligence and how technology is changing our definition of who or what an artist can be. It is about the transferal of agency onto these machines, Miller said. The artwork is saying that we are going into a period where we ask algorithms about what partner we want, what job we want, even what babies we want, end quote. Nothing more for you today. Talk to you tomorrow.