 |
|
|
|||||
Welcome to AI Decoded, Fast Companys weekly newsletter that breaks down the most important news in the world of AI. Im Mark Sullivan, a senior writer at Fast Company, covering emerging tech, AI, and tech policy. This week, Im focusing on a stunning stat showing that OpenAIs ChatGPT engages with more than a million users a week about suicidal thoughts. I also look at new Anthropic research on AI introspection, and at a Texas philosophers take on AI and morality. Sign up to receive this newsletter every week via email here. And if you have comments on this issue and/or ideas for future ones, drop me a line at sullivan@fastcompany.com, and follow me on X (formerly Twitter) @thesullivan. OpenAI’s vulnerable position OpenAI says that 0.15% of users active in a given week have conversations that include explicit indicators of potential suicidal planning or intent. Considering that ChatGPT has an estimated 700 million weekly active users, that works out to more than a million such conversations every week. That puts OpenAI in a very vulnerable position. Theres no telling how many of those users will choose their actions based on the output of a language model. Theres the case of teenager Adam Raine, who died by suicide in April after talking consistently with ChatGPT. His parents are suing OpenAI and its CEO Sam Altman, charging that their son took his life as a result of his chatbot discussions. While users feel like they can talk to a non-human entity without judgement, theres evidence that chatbots arent always good therapists. Researchers at Brown University found that AI chatbots routinely violate core mental health ethics standards, underscoring the need for legal standards and oversight as use of these tools increases. All of this helps explain OpenAIs recent moves around mental health. The company decided to make significant changes in its newest GPT-5 model based on concern about users with mental health issues. It trained the model to be less sycophantic, or less likely to constantly validate the users thoughts, even when theyre self-distructive, for example. This week the company introduced further changes. Chatbot responses to distressed users may now include links to crisis hotlines. The chatbot might reroute sensitive conversations originating to safer models. Some users might see gentle reminders to take breaks during long chat sessions. OpenAI says it tested its models responses to 1,000 challenging self-harm and suicide conversations, finding that the new GPT5 model gave 91% satisfactory answers compared to 77% for the previous GPT5 model. But those are just evals performed in a labhow well they emulate real-world conversations is anybodys guess. As OpenAI itself has said, its hard to consistently and accurately pick up on signs of a distressed user. The problem began coming to light with research showing that ChatGPT usersespecially younger onesspend a lot of time talking to the chatbot about personal matters including self-esteem issues, friend relationships, and the like. While such conversations are not the most numerous on ChatGPT, researchers say they are the lengthiest and most engaged. Anthropic shows that AI models can think about their own thoughts It may come as a surprise to some people that AI labs cannot explain, in mathematical terms, how large AI models arrive at the answers they give. Theres a whole subfield in AI safety called mechanistic interpretability dedicated to trying to look inside these models to understand how they make connections and reason. Anthropic’s Mechanistic Interpretability team has just released new research showing evidence that large language models can display introspection. That is, they can recognize their own internal thought processes, rather than just fabricate plausible-sounding answers when questioned about their reasoning. The discovery could be important for safety research. If models can accurately report on their own internal mechanisms, researchers could gain valuable insights into their reasoning processes and more effectively identify and resolve behavioral problems, Anthropic says. It also implies that an AI model might be capable of reflecting on wrong turns in its thinking that send it in unsafe directions (perhaps failing to object to a user considering self-harm). The researchers found the clearest signs of introspection in its largest and most advanced modelsClaude Opus 4 and Claude Opus 4.1suggesting that AI models’ introspective abilities are likely to become more sophisticated as the technology continues to advance. Anthropic is quick to point out that AI models dont think introspectively in the nuanced way we humans do. Despite the limitations, the observation of any introspective behavior at all goes against prevailing assumptions among AI researchers. Such progress in investigating high-level cognitive capabilities like introspection can gradually take the mystery out of AI systems and how they function. Can AIs be taught morals and values? Part of the problem of aligning AI systems with human goals and aspirations is that models cant easily be taught moral frameworks that help guide their outputs. While AI can mimic human decision-making, it cant act as a moral agent that understands the difference between right and wrong, such that it can be held accountable for its actions, says Martin Peterson, a philosophy professor at Texas A&M. AI can be observed outputting decisions and recommendations that sound similar to those humans might produce, but the way the AI reasons toward constructing them isnt very humanlike at all, Peterson adds. Humans make judgements with a sense of free will and moral responsibility, but those things cant currently be trained into AI models. In a legal sense (which may be a reflection of societys moral sense), if an AI system causes harm, the blame lies with its developers or users, not the technology itself. Peterson asserts that AI can be aligned with human values such as fairness, safety, and transparency. But, he says, its a hard science problem, and the stakes of succeeding are high. We cannot get AI to do what we want unless we can be very clear about how we should define value terms such as bias, fairness, and safety, he says, noting that even with improved training data, ambiguity in defining these concepts can lead to questionable outcomes. More AI coverage from Fast Company: Harvey, OpenAI, and the race to use AI to revolutionize Big Law The 26 words that could kill OpenAIs Sora Exclusive new data shows Google is winning the AI search wars OpenAI finalizes restructure and revises Microsoft partnership Want exclusive reporting and trend analysis on technology, business innovation, future of work, and design? Sign up for Fast Company Premium.
Category:
E-Commerce
Samsung Electronics on Thursday reported a 32.5% increase in operating profit for the third quarter, driven by rebounding demand for its computer memory chips, which the company expects will continue to grow on the back of artificial intelligence.The South Korean technology giant set a new high in quarterly revenue, which rose nearly 9% to 86 trillion won ($60.4 billion) for the July-September period, fueled by increased sales of semiconductor products and mobile phones.Samsung, which has dual strength in both components and finished products, said it expects the demand driven by AI to further expand market opportunities in coming months. SK Hynix, another major South Korean chipmaker, also reported a record operating profit of 11.4 trillion won ($8 billion) on Wednesday, which it also described as AI-related growth.Samsung’s operating profit of 12.2 trillion won ($8.6 billion) in the last quarter marked a 160% increase from the previous quarter, when it said its semiconductor earnings were weighed down by inventory value adjustments and one-off costs related to technology export restrictions on China.Samsung’s semiconductor division posted 7 trillion in operating profit for the third quarter, with the company reporting strong sales in high bandwidth memory chips, which are used to power AI applications.“The semiconductor market is expected to remain strong, driven by ongoing AI investment momentum,” the company said in a statement. The company said an advanced version of its high-bandwidth memory chips, the HBM3E, is “currently in mass production and being sold to all relevant customers,” while samples of its next-generation product, the HBM4, are being shipped to key clients. Kim Tong-Hyung, Associated Press
Category:
E-Commerce
Seven people have been arrested in the investigation of a stunning heist at the Louvre Museum in Paris, but the lavish, stolen jewels that once adorned France’s royals are still missing. In the days after the theft, a handful of experts warned that the artifacts valued at more than $100 million (88 million euros) could be melted or broken into parts. If done successfully, some say those smaller pieces could later go up for sale as part of a new necklace, earrings, or other jewelry, without turning too many heads. You dont even have to put them on a black market, you just put them in a jewelry store, said Erin Thompson, an art crime professor at the John Jay College of Criminal Justice in New York. It could be sold down the street from the Louvre.” Thompson and others say that this has become increasingly common with stolen jeweled and metal goods, noting that it’s a way thieves can try to cover their tracks and make money. It’s not like someone could publicly wear one of France’s stolen Crown Jewels and finding a market to sell the full artifacts would be incredibly difficult after everyone and their sister has seen photos of them over the last week, said Christopher Marinello, a lawyer and founder of Art Recovery International. French prosecutor Laure Beccuau made a plea Wednesday to whoever has the jewels. These jewels are now, of course, unsellable Anyone who buys them would be guilty of concealment of stolen goods, she warned. Theres still time to give them back. The jewels may be hard to monetize By breaking them apart, they will hide their theft, Marinello said, adding that these items could become even more traceless if they’re taken out of France and through jewel cutters and robust supply chains in other countries. Still, such pieces are often sold for a fraction of the value of what was stolen due to their smaller size, but also because melting or breaking down high-profile items removes the historical worth. It isn’t a simple process. The real art in an art heist isnt the stealing, its the selling, explained Robert Wittman, former senior investigator of the Federal Bureau of Investigation’s art crime team. Wittman, who has since formed his own private practice, said that the individuals behind such heists are typically better criminals or thieves than they are businessmen. Unlike others, Wittman is skeptical about the thieves successfully monetizing the artifacts they stole from the Louvre which include an emerald necklace and earrings, two crowns, two brooches, a sapphire necklace and a single earring worn by 19th-century royals. He notes the gems may still be identifiable by their clarity, for example, and gold that was refined when the pieces were made hundreds of years ago is not as pure as what’s typically in demand today. Because of what they are, theres really no point destroying them, Wittman said, while pointing to the risks of selling such high-profile stolen goods. Scott Guginsky, executive vice president of the Jewelers Security Alliance, a nonprofit trade association focused on preventing jewelry crime, also notes the age and quality of the artifacts’ diamonds. He suspects they’re probably not graded. Its not something that you can move on the open market. Its nothing that can go through an auction house,” said Guginsky, who used to run the New York Police Departments organized theft squad. Given the amount of preparation that the thieves likely put into this, Guginsky believes they have a plan for selling the jewels, even if they might first decide to sit on the jewelry and wait out suspicion. I cant see them stealing it without having an idea what they want to do, he said. Theres always a person willing to buy stolen jewelry. No matter what it is, somebody will buy it. Sara Yood, CEO and general counsel of the Jewelers Vigilance Committee, notes most jewelry businesses implement anti-money laundering programs and look out for red flags like unusual orders, repeated purchases, and requests for secrecy. Still, she and others say the age of some jewelsif broken down effectivelycould actually make it harder to track. Newer gemstones, for example, sometimes carry a laser inscription inside that can be evaluated in a lab. But because these are historical pieces, its rather unlikely that it has those identifying features, noted Yood. Experts like Thompson say bigger gems can be recut to a point that they’re unrecognizable. A challenge is finding people who have the skill to do that and don’t ask too many questions but it’s possible, she said. Whether the people behind the heist had those contacts or certain buyers lined up is unknown. But it’s important to also note that the guys who actually enter the museums are usually all hired hands, and theyre almost always caught in these cases, Thompson added. Chances of recovery look dim She and others say that museums have increasingly faced a rash of similar thefts over recent years. Thompson notes that stealing from storage can go undetected for longer: the British Museum in London, which has accused a former curator of stealing artifacts and selling them online, is still trying to recover some of the 2,000 items stolen. Some past thieves have made ransom demands for stolen artwork overall, or wait for a potential no questions asked reward from an insurance company which can amount to about a 10% cut for some insured pieces in Europe, Thompson says. The jewels stolen from the Louvre, however, were not privately insured. Sometimes government offers of a reward for information about a high-profile heist can also quicken the investigation, although the French government has yet to publicize such an incentive. If that changes, or promising leads are uncovered from the evidence left behind at the Louvre, experts like Wittman note it could increase the chances of recovering the artifacts. Still, as more time passes, others feel that the fate of finding the historic jewels looks dim. I think theyre going to catch the criminals, Marinello said. But I dont think theyll find them with the jewels intact. Wyatte Grantham-Philips and R.J. Rico, Associated Press
Category:
E-Commerce
I-P-Go! Shares of Navan, a travel-tech firm based in Silicon Valley, hit the exchanges on Thursday. The company priced its initial public offering at $25 per share, raising roughly $923 million. The $25 per-share price is within the $24$26 range the company zeroed-in on last week, when it also announced it would sell nearly 37 million shares of common stock. The IPO puts Navans valuation at around $9.2 billion. Navan shares will trade on the Nasdaq under the ticker NAVN. Founded in 2015, the company bills itself as an all-in-one business travel, payments, and expense management platform that makes travel easy for frequent travelers, helping customers find flights and hotels, automating expensing, and delivering an intuitive experience travelers love and finance teams rely on. In other words, the company utilizes AI technology to simplify corporate and business travel, aimed at reshaping an industry that has not changed in 30 years, per its SEC filings. Navan is attempting to fly as post-pandemic travel rebounds in a big way. Data from the Global Business Travel Association shows that worldwide business travel spending was expected to reach a record $1.57 trillion by the end of 2025, although that estimate may be blunted by the government shutdown and other factors. Notably, Navan is perhaps the most high-profile IPO over the past month, since the federal government shut down. The SECs filing process has been significantly stalled as the regulator has furloughed 90% of its staff. That put a damper on the overall IPO market, which had rebounded this year: During the third quarter of 2025, EY reports that 65 IPOs raised $15.7 billion, a big increase from the 40 IPOs that raised $8.6 billion during Q3 2024. All told, there have been 176 IPOs year-to-date totaling more than $30 billion, a 20% increase over 2024. SEC Chairman Paul Atkins, in a statement on X, noted that MapLight Therapeutics, a relatively small biotech firm, went public earlier this week, and that it was proof that the governments regulatory machinery was still working. With yesterdays listing of MapLight, the IPO market is still open for businesscompanies are going public during the government shutdown using the method Congress originally intended, he wrote.
Category:
E-Commerce
House flippers are about to get an AI assist on their next renovation project. Kai is an AI-powered tool that can visually identify what’s needed to fix up a house and put it on the market. The system converts photos and videos of house projects into SKU-level material specifications and cost estimates, making it fast and easy for an institutional home renovator to create an actionable renovation plan and order all the materials needed to get the job done. Kai has just launched a partnership with home improvement retailer Home Depot to link its building material and product selection tools directly with Home Depot’s 3.5 million item inventory list. Home renovators will be able to use Kai to scan a project, identify the materials needed, and immediately order them from Home Depot. [Image: Kai] Kai’s cofounder and CEO Or Agassi calls it an “AI-native renovation intelligence platform” that can effectively reverse engineer a scope of work on anything from a kitchen remodel to a down-to-the-studs gut renovation. With a guided photography process, Kai analyzes images of a house to determine what materials are needed to complete a given project. A photo showing a cracked door, for example, would lead the tool to offer up a few options for replacement, with their prices and availability at Home Depot stores in the area. An image of an old furnace can lead the tool to recommend a more efficient model that can increase the resale value of the home. [Image: Kai] It’s an evolution of RenoWalk, an in-house tool Home Depot had used for more than a decade, which allowed business customers to streamline property management and renovation projects by integrating with Home Depot’s inventory and ordering systems. The company decided to sunset that tool, and now Kai is taking its place, using data from the tens of thousands of renovation projects completed using the old system. Kai declined to disclose terms of the Home Depot deal. Kai’s first users are large institutional housing investors and operators, including both public and private companies. Citing nondisclosure agreements, Kai declined to reveal the names of these companies. Some of these companies own more than 100,000 houses across the U.S. They also bring their own wealth of renovation data, which Kai uses to train client-specific AI models. “If you send us a scope of work of how you like to renovate, we will know how to create a playbook based on that,” Agassi says. The system also adjusts for different types of renovation projects and budget levels. “Whether you’re looking to optimize for selling to the market, selling off market, renting the property out, we know how to understand the renovation condition and tell you what needed to be done to hit this target renovation,” Agassi says. [Image: Kai] Agassi says the system has the potential to drastically reduce project timelines and budgets, and help get more affordable housing on the market. “There’s a shortage of five million single family rentals out there. And it’s only getting worse because new inventory is unable to catch up,” Agassia says. “We are trying to solve for the biggest issue in housing.” Agassi has been looking at the housing challenges in the U.S. for more than a decade. Previously he and Kai cofounder Tov Arneson created Stoa, a now-closed proptech company that helped connect fix-and-flip housing investors with institutional investors and housing operators looking for more inventory. The company was seeing solid growth since its launch in 2017 but a hike in interest rates in 2023 hit the company hard and it shut down. Agassi says that even though the company failed, the idea behind Stoa stuck with him. Kai is an evolution of that concept. “We are fulfilling what used to be our dream at Stoa, which is a system that you get into a property, take some photos, and it tells you what you need to do in order to do an optimized renovation,” says Agassi. “The technology was never there, and now it is.” Agassi says Kai expects to be serving tens of thousands of customers by the end of the year, and plans on expanding the tool for use by mom-and-pop landlords, DIYers, and contractors next year. “We start with the largest operators, but the goal is to go all the way down to the consumer sitting in the comfort of their own homes, taking some photos, and we tell them exactly what’s needed to be done in order to renovate for the optimal outcome,” Agassi says.
Category:
E-Commerce
Sites : [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] next »