Tolga Akmen/ POOL EPA, Alex Brandon AP, The Conversation There was a time when Elon Musk and Sam Altman were friends. But the two tech billionaires are now embroiled in a bitter legal battle in the United States that could reshape not just OpenAI, the artificial intelligence (AI) firm behind ChatGPT they cofounded in 2015, but also the future of the technology more broadly. Launched by Musk in 2024, the lawsuit is the culmination of a years-long feud that centres on the evolution of OpenAI from a non-profit to a for-profit enterprise. The trial, which kicked off this week in California, is expected to last roughly three weeks. But its ripple effects could be felt for many years to come. The case and the cast The lawsuit pits Musk against Altman, OpenAI president Greg Brockman, OpenAI itself, and Microsoft, the AI firm’s largest backer. Musk cofounded and helped fund OpenAI to the tune of about US$44 million. By his own account from the witness stand this week, he “came up with the idea, the name, recruited the key people, taught them everything I know, provided all of the initial funding”. Brockman served as technical cofounder; Altman became chief executive in 2019. Their alliance with Musk fractured as the organisation grew. Musk departed the board in 2018. He says he was pushed out. However, OpenAI says he walked when denied majority control. Musk subsequently launched his own rival AI venture, xAI, which is now part of SpaceX. What Musk is alleging As part of the lawsuit, Musk is alleging breach of contract, breach of fiduciary duty, false advertising and unfair business practices. His core claim is that Altman and Brockman induced him to donate on the understanding that any artificial general intelligence – or AGI – built at OpenAI would stay “open” and shared with humanity. Instead, Musk argues, the founders turned the charity into a “wealth machine”. They did this in two stages. First, via a 2019 capped-profit subsidiary. Here, OpenAI’s for-profit unit limited the returns, with the excess handed back to the nonprofit. Second, through a full restructure into a public benefit corporation, which is now valued at roughly US$852 billion. Musk’s lawyers told jurors Altman and Brockman “stole a charity, full stop”. Outside court, Musk has been throwing insults at his opponents, prompting the judge to threaten a gag order. OpenAI flatly rejects Musk’s narrative. As its lead counsel, William Savitt, told jurors: We are here because Mr Musk didn’t get his way with OpenAI. The company alleges, as described in two pre-trial blog posts, that Musk himself proposed merging OpenAI with Tesla in 2017 and walked away when denied majority control. The lawsuit, OpenAI says, is “motivated by jealousy” and designed to damage a competitor. A company under pressure The trial arrives at a precarious moment for OpenAI. The New Yorker magazine recently published an investigation describing Altman as a “pathological liar”. The investigation drew on an internal dossier compiled by OpenAI’s former chief scientist Ilya Sutskever which alleged a “consistent pattern of lying” to the company’s board. Altman called the piece “incendiary” but acknowledged “a bunch of mistakes”. Musk has been amplifying the article to his X followers throughout the trial. Financially, OpenAI is bleeding. Internal projections point to roughly US$14 billion in losses for 2026 alone, with cumulative losses expected to top US$44 billion before any profit materialises. Shortly before the trial began, OpenAI quietly shut down Sora, its flagship video-generation model. Before closing, it burned around US$1 million a day in computing costs. The closure took down a US$1 billion Disney partnership with it. Even a fresh US$122 billion fundraise from Amazon, Nvidia and SoftBank has not eased the pressure. What Musk wants Musk wants the jury to unwind OpenAI’s for-profit conversion, remove Altman from the nonprofit board, and strip both Altman and Brockman of their roles in the for-profit entity. He is also demanding US$130 billion in damages from OpenAI – for what his team calls “ill-gotten gains”. He has accused Microsoft of “aiding and abetting” and argues it is liable for a share. His legal team argues OpenAI’s existing models already constitute AGI, because they have surpassed human intelligence in many tasks. Under the founding agreement, AGI could not be commercially licensed. This would include the licence currently used by Microsoft for CoPilot. What’s at stake If Musk wins, the consequences would be significant. OpenAI’s planned initial public offering would almost certainly be derailed. This is expected in late 2026 at a US$1 trillion valuation. Investors in the recent funding round could face clawbacks. Altman, the public face of the AI boom, could be removed from the company he has led since 2019. The broader question of whether AI labs founded as charities can lawfully pivot into commercial enterprises would be settled, at least in California. This has potential implications for Anthropic and other mission-driven peers. Even a defeat for Musk would not end the controversy. The trial has already pried open Silicon Valley’s normally sealed boardrooms, surfacing diaries, Slack threads and HR memos that paint an unflattering portrait of OpenAI’s governance. The case crystallises a wider public anxiety: an incredibly powerful technology is being built and controlled by a tiny number of feuding tech bros. And it’s the rest of us who have to live with the consequences. Rob Nicholls is a part of the University of Sydney Centre for AI, Trust, and Governance and receives funding from the Australian Research Council.

Discussion | Link

Wen-Yee Lee / Reuters: Filing: TSMC sold its remaining 1.11M Arm shares at $207.65 per share for a total of ~$231M this week; TSMC invested ~$100M in Arm's 2023 IPO at $51 per share  —  Taiwan Semiconductor Manufacturing Co (2330.TW), the world's largest contract chipmaker, has sold its remaining stake in Arm Holdings …

Fabrice Coffrini / Getty Images Earlier this month, multibillion-dollar US tech company Palantir posted on X a summary of its chief executive Alex Karp’s recent book, the portentously titled The Technological Republic: Hard Power, Soft Belief, and the Future of the West. The book and the post offer a kind of manifesto, making sweeping claims about a hierarchy of civilisations, the rejection of pluralism, Silicon Valley’s moral obligation to US military power, the necessity of AI-powered weapons, and the case for compulsory military service. The manifesto has met widespread criticism. Some commentators have compared the rhetoric to the monologue of a comic-book villain: grand, moralising, tinged with a sense of historical destiny. But the manifesto is more than just corporate posturing: it’s helping to construct a new geopolitical reality and normalise a worldview that concentrates power beyond democratic accountability. From tools to worldviews For the past two decades, large technology firms have mostly presented themselves as benevolent service providers. They build tools; governments and users decide what to do with them. That distinction has always been convenient, but it is looking less and less tenable. For some, Karp’s manifesto offered a grim sense of confirmation of the change. As Austrian philosopher Mark Coeckelbergh put it, “reading it is like opening a food item that you suspected has gone off, but you didn’t know it was that much off”. Palantir is not just any tech company. Its software, offering “AI-powered automation for every decision”, is embedded in military, intelligence and policing systems – not just in the United States, but in many other countries across Europe, the Middle East and Australia. When a company in that position denounces “regressive” cultures and “hollow” pluralism, it is asserting a worldview rather than just selling technology. As the manifesto puts it: “the ability of free and democratic societies to prevail requires something more than moral appeal. It requires hard power.” Here, “hard power” means not just military force but the technological systems that shape how force is used. Palantir’s systems shape how threats are identified, interpreted and acted upon. So when the company advances claims about civilisational decline and the necessity of strength, it is also helping define the terms on which power is exercised. A stakeholder letter or something older? In one view, the manifesto is a corporate position paper or a statement of values aimed at investors, partners, the public and policymakers. But there is something older in its form. It is reminiscent of Cicero, the Roman statesman and master of rhetoric, in its talk of decline, virtue, duty and the survival of the republic. It frames technological development not as a market activity but as a moral obligation tied to the fate of civilisation. Like classical republican oratory, it asserts that survival depends on strength. And today, that strength is technological. Cicero wasn’t simply expressing his own opinions when he spoke. He was asserting a right to speak on behalf of the republic. In the same way, Palantir is positioning itself as a legitimate interpreter of civilisational stakes. The shift from argument to atmosphere The manifesto does not argue via carefully reasoned policy claims. Instead it offers declarative statements: that some cultures are “harmful”, that pluralism has become “vacant”, that technological strength is the ultimate guarantor of civilisation. These establish a mood: urgency, decline, necessity. The effect is to manufacture a sense of inevitability. It works via tone and framing rather than evidence, setting the background conditions under which certain policies feel necessary rather than debatable. Once that atmosphere is in place, the range of acceptable responses shrinks. Palantir is helping to construct geopolitical realities, rather than respond to them. Supervillain or Cicero? It’s both Palantir’s rhetoric does bear comparison to the ranting of fictional supervillains. Both feature sweeping claims about decline and the need for decisive action. Palantir also exempts itself from the accountability that might accompany its claims. Comic-book villains believe they see more clearly than others, but they also place themselves above constraints that apply to everyone else. The structure of the argument feels familiar. The world is in crisis, the options are narrowing, and power must be expanded beyond normal limits. Seen this way, the villain tone and the Cicero-like register are two expressions of the same underlying move. It is an effort to define reality at a civilisational scale, from a position that answers to no one. An infrastructure project This worldview did not emerge overnight. It has been developed over years through op-eds in prestige newspapers and published by major mainstream houses before being compressed into a social media thread that reached millions in hours. When companies that build and operate core security technologies put considerable resources into developing and promoting stories about civilisation and its future, their language is not just expression. It is a kind of infrastructure for their actions in the real world. By the time most people notice the rhetoric, the infrastructure it justifies is already in place. But the future trajectory of this worldview is not set. The history of democratic politics is, in part, a history of people recognising when power has overreached and building the collective capacity to say so. That work is not heroic in the comic-book sense. It doesn’t focus on a single figure or decisive moment. It starts with understanding precisely how the manufacture of inevitability works, so what is presented as necessary can be seen as a choice – before it is made for us. Daniel Baldino does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

Tory Burch says she didn't want to compromise on family time even as her company demanded more of her attention. ANGELA WEISS / AFP Tory Burch says she would "never trade off" being a good mom for her career. If anything, her social life suffered — and she gave up sleep, she said. "But I will say, being a mom is the joy of my life," the fashion designer said. Tory Burch, 59, says her career success didn't come at the expense of being a mom — but it did cost her a social life. Speaking on Tuesday's episode of the "Aspire with Emma Grede" podcast, the fashion designer said she refused to compromise on family time, even as her company demanded more of her. "I mean, one thing I would never trade off was being a good mom," Burch told podcast host Emma Grede. "I never thought I could start a business if that were to have to happen." Burch shares three sons with her ex-husband, J. Christopher Burch, with whom she co-founded her namesake label in 2004. The pair divorced in 2006. "When we got divorced, I was raising my kids a lot on my own. So, it was being at the doctor, going to a lacrosse game," Burch said. Drawing from her own experience, Burch said she set out to build a flexible company culture that supports women balancing work and motherhood. "I have done it, and what I say is that as long as it gets done, I'm not really clocking how it gets done," Burch said. She also made a point of leaving work as soon as she could. "And I used to do that, maybe not at five, but I would be a stickler for being at home," she said. "I would say, if anything suffered, it was my social life. She said that was "fine" with her, since motherhood has been the most fulfilling part of her life. "But I will say, being a mom is the joy of my life. And I often say I'm a survivor of raising three boys, because it's hard," she added. Burch recalled how her sons were like little "torturers" who used to "bully" her, though she was a "tough mom." Now in their 20s, they sometimes apologize for what they put her through. Looking back, Burch said she didn't see it as making sacrifices, but as working around both her personal and professional responsibilities. "I guess the only thing that was sacrificed is my sleep," she said. "I still don't sleep, but I think, you know, for me, I would, you know, get the kids handled and put to bed and then start work." It was simply part of the reality of running a company, Burch added. "But at the same time, I will say I can turn it off very easily, too. I have a lot of outside passions." Burch isn't the only high-profile figure to speak about the realities of balancing work and motherhood. In October, Shonda Rhimes said it was a relief to accept she couldn't excel at her career and be the perfect mother at the same time. "It's never going to happen. You're never going to feel amazing at both. It's always going to be a little bit, you know, shaky. So let it be shaky," she said. In April, Rebecca Minkoff told Business Insider she doesn't believe in work-life balance — and said she struggled to do it all while raising her first three children. "I wish someone had sat me down and shaken me and been like, 'Only your babies matter, focus on them,'" Minkoff said. Read the original article on Business Insider

Bloomberg: Sources: China suspends issuing Level 4 autonomous vehicle licenses, after 100+ of Baidu's Apollo Go robotaxis stalled and disrupted traffic in Wuhan in March  —  China has suspended issuing new licenses for autonomous vehicles, according to people familiar with the matter …

Hilton's CEO says he sees the K-shaped economy converging. Justin Sullivan/Getty Images Hilton's CEO says he sees the K-shaped economy converging into a C-shape. He said macro trends are boosting spending from lower-income consumers. However, other company executives have said that the economy is more divided than ever. Hilton's CEO said America's K-shaped economy is converging as lower-income consumers are spending more. During a Tuesday earnings call, CEO Christopher Nassetta said that he expects "improving performance in the lower and mid-chain scales" in the rest of the year. He said sales growth would continue to "move downstream from luxury and upper upscale toward a more balanced convergence demand shape, or what I have been calling a C-shaped economy." The term is a play on the K-shaped economy. Analysts and company executives have warned about the economy going K-shaped in recent months — a difference in spending behavior between high- and low-earners. When asked why he predicted a "C-shaped economy," Nassetta said factors like falling inflation, the expectation of lowered interest rates, and heavy investment in AI are "benefiting the middle and lower income consumer and driving broader demand growth." He said these were broad, big-picture trends, adding that people should "forget, for the moment, the spike in energy prices and oil because of the war in Iran." Hilton reported a 3.6% increase in RevPAR — a measure of revenue per available room — in the first quarter of 2026 compared to the same period last year. Nassetta's comments are at odds with those of other industry executives, who have warned that the K-shaped economy is more divided than ever. Executives of Delta Air Lines, Macy's Inc, and Ralph Lauren previously said that their affluent customer base has driven sales. On the other side of the "K-shape," food and beverage brands are slashing prices to attract lower-income spenders. PepsiCo's CEO Ramon Laguarta announced in February that the company would cut the prices of some of its products, such as Lay's, Doritos, and Cheetos, by up to 15%. And at the start of April, McDonald's announced that it was expanding its value menu to include $3 items and $4 meal deals. However, Visa's CFO on a Wednesday earnings call gave promising indicators that consumer spending is holding strong. Visa finance chief Christopher Suh said the US payment volume grew 8% in the latest quarter compared to the same period the year before, "reflecting resilience and consumer spending." Read the original article on Business Insider

Nikkei Asia: Sources: Apple tells suppliers it will maintain its current production share in India rather than expand, after local sites struggled without China-based teams  —  PALO ALTO, California/TAIPEI — Apple's incoming CEO has a China problem: On the one hand, the iPhone giant is finally seeing sales rebound …

Seema Mody / CNBC: Pentagon Digital and AI Chief Cameron Stanley confirms the DoD is expanding its Google Gemini use, saying “overreliance on one vendor is never a good thing”  —  Pentagon AI chief Cameron Stanley confirmed to CNBC that the Department of Defense is expanding its use …

An artisanal miner holds a cobalt stone at a mine near Kolwezi, Congo, in 2022. About 20,000 people work there among toxic materials. Junior Kannah/AFP via Getty Images There is a troubling contradiction at the heart of the global transition to a cleaner, greener, tech-driven future: Modern technologies – everything from AI to wind turbines, as well as cellphones, electric vehicles and defense systems – depend on critical minerals. But many of the communities where those minerals are mined end up with polluted water and poorer health because of the mining. Lithium powers batteries. Cobalt stabilizes them. Copper carries electricity. Rare earth elements make wind turbines and digital devices efficient and durable. Each of these are essential to the technologies of the fourth industrial revolution, but they are also toxic and require enormous amounts of water to extract. As researchers at the United Nations University Institute for Water, Environment and Health, we have been studying the impacts of critical mineral mining on communities around the world. Our new report shows why mining will end up worsening the lives of some of the world’s poorest people if critical mineral supply chains are not monitored and regulated. One of us is from the Middle East, a region still suffering from the long-term consequences of supplying the fuel consumed for the remarkable economic developments of the 20th century. And one of us comes from Africa, the continent that is now serving as a major supplier of the critical minerals that fuel technological advancements in the 21st century. Based on our experiences and our research, we believe that if there aren’t major changes in how countries, corporations and communities manage critical minerals, humanity risks reproducing the injustices of the oil extraction era, this time with the technological advancements meant to address the problems fossil fuels created. Mining contributes to growing water bankruptcy One of the most significant impacts of critical minerals extraction is its effect on water. In 2024 alone, global lithium production required an estimated 456 billion liters of water. That is equivalent to the annual domestic water needs of roughly 62 million people in sub‑Saharan Africa. At the same time, much of the world is facing water bankruptcy, meaning people and industries are using more fresh water than nature can replenish, leading to irrecoverable ecosystem damages. Workers perform maintenance at pools where evaporation concentrates lithium-rich brine in Chile’s Atacama Desert in 2023. To extract lithium, mines pump water from beneath the salt flats. AP Photo/Rodrigo Abd In arid regions such as Chile’s Salar de Atacama, mining activities account for up to 65% of total regional water use, competing with agriculture and ecosystems. Groundwater levels have dropped, salt lagoons have shrunk, and freshwater aquifers are increasingly at risk of being depleted and contaminated. Water pollution compounds problems like this. Mining generates large quantities of toxic waste and wastewater containing heavy metals, acids and radioactive residues. Source: United Nations University Institute for Water, Environment and Health Rare earth mineral production, for example, generates up to 2,000 metric tons of waste for every metric ton of usable material. Rare earth minerals are often extracted by creating leaching ponds and adding chemicals to separate the metals. When the effluent isn’t treated or is improperly stored, the chemicals can seep into groundwater and waterways, contaminating aquifers and rivers. In some parts of the world, rivers near cobalt and copper mines have become so acidic that communities can no longer drink water from them. Fish stocks have collapsed, and farmlands have been poisoned. Water insecurity is no longer a side effect of mining; it is a systemic cost. Health crises hidden in supply chains Communities living near these extraction sites report people suffering from skin diseases, gastrointestinal illnesses, reproductive health problems and chronic health conditions associated with long‑term exposure to heavy metals in polluted water and soil. Evidence from mining regions in the Democratic Republic of the Congo is particularly stark. Studies document high rates of miscarriages, congenital malformations and infant mortality among populations exposed to environments contaminated with cobalt and other metals. Maternity wards in southern Democratic Republic of the Congo that are close to mining operations report significantly more birth defects than those farther away. In communities near mining operations, residents talk about how women and girls living near cobalt and copper mining sites have been experiencing gynecological health problems, including infections, menstrual irregularities, miscarriages and infertility. These risks are linked to prolonged contact with contaminated water, compounded by limited access to sanitation and healthcare. In Chile’s Antofagasta region, cancer mortality is the highest in the country. Lung cancer rates there are nearly three times the national average. Physicians in the region also report rising cases of neurological and developmental disorders, which they link to early exposure to contaminated water and air. Thousands of children are estimated to be employed in artisanal cobalt mines in the Democratic Republic of the Congo. In the informal mines, they may be exposed to cobalt dust and other hazardous materials without protective gear. These health risks are heightened by weak systems for water, sanitation and healthcare. As of 2024, only about one-third of people in the Democratic Republic of the Congo had at least basic drinking water services. Food costs of the energy transition The water problems caused by critical minerals extraction also pose a major threat to local food systems. In Peru, zinc mining has contaminated the Cunas watershed. Runoff pollutes water used to irrigate crops and provide water for livestock. In Bolivia’s Uyuni region, lithium mining has led to persistent water shortages that are making it increasingly difficult to grow quinoa, a staple crop central to local diets and economies. Across the wider “lithium triangle” of Argentina, Chile and Bolivia, mining has reduced water availability for crops and farm animals. Similar patterns are evident in parts of the Democratic Republic of the Congo and Zambia. In both countries, polluted rivers have contributed to declining fish stocks and livestock illnesses, harming households that are already struggling to feed themselves. Ways to protect mining communities Innovation and technological advances have the potential to do good. But we believe a fair and sustainable energy and digital transition requires deliberate actions to avoid creating “sacrifice zones,” places where human and ecological well-being are traded away for technological breakthroughs. A family works at an artisanal cobalt and copper mine site in 2025 in Kolwezi, Democratic Republic of the Congo. These mines are often unregulated. Michel Lunanga/Getty Images One option is to create stronger international governance. Moving beyond voluntary guidelines toward binding international rules, such as treaties, enforceable supply chain due-diligence laws, mandatory environmental and human rights standards for mining operations, and potentially establishing a global mineral trust that would manage critical minerals as shared planetary assets, could improve water protection, pollution control and human rights across mineral supply chains. Companies can also invest in less water-intensive mining technologies. Countries can tighten their wastewater controls and expand independent environmental monitoring and reporting. Copper-mining companies create huge tailings ponds, like this one in Chile in 2019, to store toxic byproducts of mining. Hundreds of these waste ponds exist across the country and carry the risk of leaking acidic water and heavy metals such as arsenic, copper and mercury into groundwater. Martin Bernetti/AFP via Getty Images Governance arrangements that give local and Indigenous communities a stronger voice, a fair share in the benefits and genuine co-governance of resources could further rebalance who has power and who bears risk. On the consumption side, extending product lifespans, expanding recycling and encouraging less reliance on newly mined minerals would ease pressure on water‑stressed regions. For the people who use these technologies, the social and environmental costs embedded in critical minerals supply chains are often out of sight and out of mind. Making these impacts visible can enable consumers to make informed choices and engage in greater scrutiny of corporate practices. Critical minerals are essential to advancing sustainability. But if cleaner technologies are built in ways that result in polluted rivers, sick children and dispossessed communities, the transition will fall short of its promise. The authors do not work for, consult, own shares in or receive funding from any company or organization that would benefit from this article, and have disclosed no relevant affiliations beyond their academic appointment.

From an 8-year-old to a retiree, these builders have redefined who gets to create with AI. Boo Kok Chuon; Lee Chong Ming/Business Insider; Lewis Dickson What began with developers has spilled into playpens, classrooms, and retirement homes. It's not just non-technical adults who vibe code now — it's grandparents and kids still learning to read. Most of them told Business Insider they don't read — or understand — the code. They saw what AI tools could do, got intrigued, and started tinkering. Vibe coding refers to the practice of using plain English to tell AI what to build. It has taken off since Andrej Karpathy last year urged developers to "fully give in to the vibes." For Lena Hall, a senior director of developer and AI strategy at Akamai, that meant handing the mic to her 5-year-old. Using voice mode, her son built a game in about 20 minutes with OpenAI's Codex. Others, like 13-year-old Usman Asif, started vibe coding after learning about it in a workshop. The process was bug-filled. 13-year-old student Usman Asif learned vibe coding and competed in Cursor's 24-hour hackathon in Singapore. Lee Chong Ming/Business Insider "It kind of drove me crazy because I did not know what to do," the teen said. "But it's like that, you know, one bug after another, then you get there," he added. Asif has joined hackathons, built an AI sports coach, and co-created an AI-powered university guidance counselor. At the other end of the spectrum, Lewis Dickson, a 78-year-old retiree, said he picked up vibe-coding when the tools went mainstream. Lewis Dickson says being retired doesn't mean he can't use AI like young pros. Lewis Dickson The tech enthusiast told Business Insider that people often think "gray hair means outdated technology skills." "I understand where that perception comes from, but it's not always accurate," he added. Read the stories of the unlikely coders changing who tech is for Read the original article on Business Insider

Rei Onoda returned to Japan to get married, then she and her husband left Tokyo to open a bakery in the countryside. Provided by Rei Onoda Rei Onoda was a flight attendant based in Dubai for four years. She moved back to Japan and left Tokyo for the countryside while pregnant. Now, she runs a bakery from a 180-year-old home with her husband and their toddler son. This as-told-to essay is based on a conversation with Rei Onoda, 34, coowner of Kamijyoan, a bakery in rural Japan. Her words have been edited for length and clarity. After graduating from university, I moved to Dubai and worked as a cabin crew member for Emirates for about four years. I got to travel all over the world. Seeing so many places made me realize Japan was actually my favorite. Being away made me appreciate how unique it is. I was also in a long-distance relationship at the time. After we got married, I moved back and decided to change jobs. I wanted a new challenge and to use my English skills in a different way. A shift back to Japan I joined a PR agency in Tokyo and worked on media tours across Japan, bringing people from overseas not just to big cities but also to local areas. That's when I realized how many beautiful areas in Japan I didn't even know about. I started thinking about living somewhere more rural. When COVID-19 started, my job became fully remote. I lived near Shibuya Station, a crowded part of Tokyo, but everything was closed, and I was stuck at home paying high rent. The house sits on about 6 acres of farmland and requires a registered farmer to manage the land. Provided by Rei Onoda Discovering rural life My husband and I started talking about moving to a quieter place. We had always imagined doing that later in life, but we thought: Why not now? His hometown is about an hour and a half from Tokyo, so we began looking for a house nearby, expecting it to be a temporary move. The house in the Japanese countryside had been empty for 10 years. Provided by Rei Onoda We weren't planning to open a bakery — just to find a place to live. After about two years of searching, we bought a 180-year-old house in Kamijo Village, about 140 miles northwest of Tokyo, for 8,500,000 Japanese yen, or about $50,000. The purchase included 6 acres of farmland. The house had been empty for about 10 years and wasn't livable without major renovation. Because farmland is strictly regulated in Japan, my husband — who had been working in fashion retail — agreed to become a registered farmer to manage the land. When they bought the house, it was cold, and the tatami was damaged. Provided by Rei Onoda We found that to renovate the property as a home, we would have to follow strict preservation rules. That would have been too expensive, so we started thinking: Instead of living there, maybe we could open a small café and share the space with others. That was the starting point of our bakery. It wasn't as simple as we thought Because the house is in a preservation district, we needed approval from the city, the prefecture, and the national government. The entire project, including renovation, took about three years. We funded it through savings, bank loans, and grants, spending about 50 million Japanese yen, including kitchen equipment. We also had to preserve the building's historical value while making it safe, working with architects and Japan's Agency for Cultural Affairs. We received 29 million Japanese yen in grants for regional development and cultural preservation by positioning the project as both a café and a means to preserve a historic property. She positioned the project as both a café and a means to preserve a historic property. Provided by Rei Onoda Building relationships with the local community was challenging Kamijo Village is very small, and many residents have lived there for generations. When we bought the house, we went door-to-door to introduce ourselves. Even while still living in Tokyo, we visited often, joined local events, and helped with community activities. Eventually, we moved there before the project was finished. It wasn't easy to be accepted as newcomers, but by showing up consistently, we slowly built trust. Onoda and her son in the Japanese countryside. Provided by Rei Onoda During those three years, there was also a lot of waiting I didn't want to waste that time, so I went to baking school in Tokyo and trained with a baker who had worked in Germany. I was also pregnant at the time. It wasn't as difficult to juggle as I'd expected — most of my work was paperwork, like applications and funding preparation, which I could do from home. I was excited about both the construction and becoming a mother. Now, I'm in charge of the bakery, and my husband handles the coffee. Onoda's husband REWRITE Provided by Rei Onoda We opened about a year and a half ago. At first, most customers were local. After about 6 months, more tourists started coming, finding out about us through Instagram and Google Maps. On a typical day, I bake a handful of loaves — country-style, sandwich bread, and rye — and prepare cakes and cookies at the start of each week. We usually see around 15 groups on weekdays and closer to 25 on weekends. Balancing the bakery and motherhood has been challenging. Onoda found balancing the bakery and motherhood challenging. Provided by Rei Onoda At the same time, I learned to rely on others. With support from my husband's parents and daycare, I began to understand the importance of asking for help. It was not something I was good at before, but I now feel that being able to ask for help is essential — not only for raising a child, but also for running a business. We moved into a house near the bakery. This part of Japan is known for growing fruits and wine. We've started making a craft peach beer using local fruit. Onoda and her husband have learned to work well together. Provided by Rei Onoda I still use what I learned as a cabin crew member Being a cabin crew member taught me how to work efficiently in a small space and think about the customer experience. I use those skills in my bakery every day. If I give advice to someone who wants to do something similar, it's important to clearly explain why you're doing it and what value it has. Starting a business in a rural area takes more time than you expect. But that time wasn't wasted — it became the foundation. This bakery exists because of the support we received from the local community. I hope that as my son grows up, he'll see that and understand the value of building something with others and not giving up on what you believe in. Read the original article on Business Insider

Red Lobster brought back its Endless Shrimp promotion on Monday after blaming the all-you-can-eat crustacean deal for its bankruptcy. 4kodiak/Getty Images/iStockphoto Red Lobster once blamed its blockbuster Endless Shrimp deal for its downfall. Now the all-you-can-eat promo is back — but the vibe is different. Some customers worry the value isn't the same, and the deal isn't driving the traffic it once did. Endless Shrimp is back — but it's not exactly causing a stampede. On a recent evening at a Red Lobster in Los Angeles, a steady stream of diners ordered the all-you-can-eat deal, and the restaurant was busier than usual. However, large sections of the dining room sat empty, and the pace never tipped into the chaos that servers remember from past runs of the promotion. "It's been popular," one server said, "but not as popular as I would have thought." That's a notable shift for a deal that once defined the brand — and, at times, overwhelmed it. Red Lobster has brought back Endless Shrimp for a limited run, six weeks in Los Angeles, servers said, reviving a promotion it once said it couldn't afford to keep. CEO Damola Adamolekun told Business Insider last year that he had "no plans" to bring it back, after the company cited the deal as a key factor in its 2024 bankruptcy. Now, at a time when the business could use a boost, Red Lobster says it's reviving Endless Shrimp in response to customer demand. New data suggests a more muted comeback. Every table in the occupied sections of a Los Angeles Red Lobster had a menu display promoting the return of Endless Shrimp. Katherine Tangalakis-Lippert Foot-traffic data from Advan shows Red Lobster saw a clear surge in visits during a 2023 Endless Shrimp push, when it first made the promotion a permanent menu fixture. That bump wasn't seen in 2024 or 2025, after it was removed from menus, and hasn't been as significant this time around. Advan's data shows that foot traffic the week of the latest Endless Shrimp launch was down 0.9%, well below the prior surge and a dip compared to positive trends in March. More recently, traffic has been uneven, according to analysis by Placer.ai. After seeing a five-month surge last summer, when Adamolekun launched the chain's viral seafood boils, visits have been down year over year since November 2025. Placer did not have data available regarding the performance of the Endless Shrimp relaunch at the time of publication. The chain's 2026 performance has been affected by "unfavorable weather, rising gas prices, and broader macroeconomic concerns," Placer's head of analytical research, R.J. Hottovy, said. He added that the shrimp promotion "should resonate with value-conscious consumers," but it's arriving in a softer, less predictable environment. Casual dining chains like Red Lobster are trying to win back customers in a tougher environment than they faced a decade ago. The industry is navigating a K-shaped economy, in which higher-income diners are still spending, but middle- and lower-income consumers — long the backbone of chains like Red Lobster — are pulling back or trading down. That's put pressure on legacy brands with higher price points, especially as competitors like Chili's lean into value deals and aggressively market cheaper options. For Red Lobster, which typically charges more than many casual-dining peers, that dynamic makes it harder to convince diners they're getting a deal — even with an "endless" one on the menu. Two former Red Lobster executives told Business Insider that Endless Shrimp worked best as a tightly controlled, seasonal promotion — something rare enough to feel like an event and strategically timed to fill slower periods. Over time, they said, the company leaned on it more heavily to drive traffic, eroding the sense of urgency that once made it so effective. "It lost its appeal," one former executive said. That shift shows up not just in the data, but in the dining room. Some customers are showing up for Endless Shrimp. One Los Angeles diner, Jeffrey Turney, told Business Insider he visits Red Lobster every couple of months when promotions are running and was "excited for the return" of the bottomless deal, enthusiastically texting friends about it as he waited for his meal. After dinner, he told Business Insider that overall the meal was "pretty good," but he "really disliked" the new Marry Me Shrimp, a social-media-driven flavor launched last week. Red Lobster's Endless Shrimp menu features five variations of shrimp dishes, including coconut shrimp and linguini Alfredo. Katherine Tangalakis-Lippert About a third of the diners at the Los Angeles location had purchased the deal during a recent dinner service, a server told Business Insider. Most skipped the bottomless shrimp, with some opting for different deals — like the $30 steak and lobster Wednesday night special, which carries the same price tag as the Endless Shrimp promotion in the Los Angeles market. "I love that they brought it back," said Destiny Howard, a social media creator and Red Lobster fan who runs the account "Munch with Des," adding that the move shows the company is "listening to their customers." She framed the appeal in terms of responsiveness and novelty, not necessarily value. For some diners, the math doesn't add up. With restaurants charging between $24.99 and $29.99 for unlimited servings of five dishes, depending on location, some felt it wasn't a particularly good bargain — especially compared with other menu options at the same price point. "I wasn't too happy with the raise in price," Turney said. "For $30, I think you should get at least two sides," he added, but noted that groceries and dining out have gotten more expensive across the board. That perception lines up with broader industry pressures: Seafood costs have risen, and casual dining chains are increasingly competing on value with both fast-food and fast-casual brands. Endless Shrimp was never just about cheap seafood. It was about drawing people in during slow months, ideally to order something with a higher margin, the former executives told Business Insider. These days, while the promotion is bringing some diners through the door, it's not clear whether it's delivering the same kind of lift. For Red Lobster, that may be beside the point. The chain is running Endless Shrimp at the tail end of its fiscal year and during a typically slower stretch, a window when a modest bump in traffic can matter. As one former executive put it, the promotion can still work as a lever to "drive business quickly," even if it no longer carries the same magic. For customers like Turney, the comeback doesn't quite hit as it used to. He said he wouldn't be back again for Endless Shrimp — but he'd be willing to give something else a go. Read the original article on Business Insider

Charles Yang has only worked in startups in China and in the US. Charles Yang Charles Yang spent 10 years building a startup in China before moving to the US to start again. He said that China's startup scene feels less structured than in the US. In China, government direction, cultural norms, and community-first values shape how startups are built. This as-told-to essay is based on a conversation with Charles Yang, the founder and CEO of Vibe, a US-based AI workspace startup. The conversation has been edited for length and clarity. I started my first company with my friend after finishing my third year at Zhejiang University. I still remember the morning in January 2006. We went to the bank, and my father gave me about $5,000. My friends and I used that sum to start our own gaming company. As a computer science major, there were conventional pathways laid out: You can go to big companies like Microsoft, Baidu, or Alibaba; pursue a master's degree, then a Ph.D, or go to the US to study for a Ph.D and stay there. I quit college after three years to build my startup. My family did care about whether I would graduate, because I was the only one in my family who attended college. I was so young that I do not think there was any concept of "risk" in my head. Moving to the US for a global career After about 10 years, my gaming company became very successful. We considered selling or going public. I was married and had a daughter, so I began thinking about my next career and life goal. If you go public, you're tied to the company for at least six years. I wanted a better answer if my kid asked what I do. I do not want to build another Chinese game. I want to be a big global player. In 2016, I moved to Seattle, met friends at Apple and Google, and saw Silicon Valley. I had spent 11 years building virtual things, and I decided to build a new startup and something real — hardware. I stepped away from my gaming company in China. I built an early-stage immersive collaboration platform exploring spatial computing for teamwork in the US, and raised $1 million in seed funding in 2016. I saw advantages in linking China's supply chain to the US ecosystem. We've managed to build products and scale the business through e-commerce, and set up distribution and sales pipelines. In 2024, I pivoted my startup to the AI workspace, and we are backed by Sequoia Capital China, now known as HongShan. In China, startups feel unstructured. In the US, they run off a playbook The US startup environment feels like a well-documented software development kit. You follow the playbook, and even some wildcards are documented in a systematic way. In China, the startup environment is more like a badly documented system. There are a lot of bugs, a lot of holes. Chinese society follows blurred, vague, and hidden rules. Usually in China, relationships are built through dinners and drinking wine with people. A lot of opportunities come from that. In China, you don't put yourself first. You are part of the community first, then yourself. In the US, it is more individualistic. The pre-pandemic period was probably the golden age for many Chinese tech startups, where many young people chose to start companies. But now, in China, resources are being allocated more to state-owned organizations and startups that prefer the government-led direction. For example, if Beijing wants something, then a lot of resources go that way. In the US, it is more organic. Advice to startup founders For someone more independent or ambitious, I would probably suggest they join a successful company for at least two or three years. I think I missed out on that at the beginning. As for when to build a company for yourself, if you have a strong interest in something, just do it. Being young and not having experience does not necessarily prevent success. At the end of the day, it is really about your passion — and having luck. Do you have a story to share about a tech career in China? Contact this reporter at cmlee@businessinsider.com. Read the original article on Business Insider

arXiv:2604.24804v1 Announce Type: new Abstract: Offline preference optimization methods, such as Direct Preference Optimization (DPO), offer significant advantages in aligning Large Language Models (LLMs) with human values. However, achieving optimal performance with these methods typically involves additional hyperparameter tuning, resulting in substantial time overhead. Although prior work has proposed a range of improvements, these methods remain limited in effectiveness and have not fully eliminated reliance on hyperparameter tuning. In this work, we propose RMiPO, a lightweight and efficient framework for offline preference optimization. RMiPO leverages intrinsic Response-level Mutual information for Preference Optimization with hyperparameter modulation, dynamically decoupling preference contributions at negligible additional computational cost. Extensive experimental results demonstrate that RMiPO achieves consistently superior performance over existing methods while reducing training overhead by more than 15\%. Our code is available at https://github.com/liavonpenn/rmipo.

OpenAI outlines a five-part action plan for strengthening cybersecurity in the Intelligence Age, focused on democratizing AI-powered cyber defense and protecting critical systems.

arXiv:2604.24919v1 Announce Type: new Abstract: Earth Observation (EO) is moving beyond static prediction toward multi-step analytical workflows that require coordinated reasoning over data, tools, and geospatial state. While foundation models and vision-language models have expanded representation learning and language-grounded interaction for remote sensing, and agentic AI has demonstrated long-horizon reasoning and external tool use, EO is not a straightforward extension of generic agentic AI. EO workflows operate over georeferenced, multi-modal, and temporally structured data, where operations such as reprojection, resampling, compositing, and aggregation actively transform the underlying state and can constrain subsequent analysis. As a result, errors may propagate silently across steps, and correctness depends not only on internal coherence, but also on geospatial consistency, temporally valid comparisons, and physical validity. This position paper argues that these challenges are structural rather than incidental. We identify the implicit assumptions commonly made in generic agentic models, analyze how they break in geospatial workflows, and characterize the resulting failure modes in multi-step EO pipelines. We then outline design principles for EO-native agents centered on structured geospatial state, tool-aware reasoning, verifier-guided execution, and learning objectives aligned with geospatial and physical validity. Finally, we present research directions spanning EO-specific benchmarks, hybrid supervised and reinforcement learning, constrained self-improvement, and trajectory-level evaluation beyond final-answer accuracy. Building reliable geospatial agents therefore requires rethinking agent design around the physical, geospatial, and workflow constraints that govern EO analysis.

arXiv:2604.24788v1 Announce Type: new Abstract: Natural gas is undoubtedly an essential component of the global energy system. Accurate short-term forecasting of natural gas price is challenging due to pronounced volatility driven by seasonal demand patterns, geopolitical developments, and shifting macroeconomic conditions. The nonlinear dynamics and frequent regime changes can limit the effectiveness of traditional time-series models. In this study, we explore the use of Liquid Neural Networks (LNNs) for short-horizon forecasting of the Henry Hub spot price, a primary benchmark for pricing. LNNs are designed to adapt continuously to evolving temporal patterns through dynamic internal state updates, making them well suited for nonstationary price behavior. By improving forecast accuracy in volatile market conditions, this work aims to reduce uncertainty and enhance decision support across energy trading and power market applications.

arXiv:2604.24768v1 Announce Type: new Abstract: In this chapter, we investigate the bending behavior of a perforated nanobeam subjected to sinusoidal loading using an efficient and computationally robust Physics-Informed Functional Link Constrained Framework with Domain Mapping (DFL-TFC) method. Our aim is to determine the relationship between static bending response and dynamic deflection of a perforated nanobeam for various perforation cases. The static bending is obtained using the FL-TFC with Domain mapped method, whereas dynamic deflection is determined using the Galerkin method. The proposed approach employs the theory of functional connections (TFC) to systematically embed governing differential equation constraints into a constrained expression (CE), which exactly satisfies all prescribed initial and boundary conditions (ICs and BCs) and domain of differential equation is mapped to domain of orthogonal polynomials. Within this framework, the free function appearing in the constrained expression is expressed through a functional link neural network (FLNN). The cost is minimized by the mean square residual of DE, allowing training without requiring complex deep network architectures. Relationship between static and dynamic defection of simply-supported (S-S) perforated nanobeams has been investigated here. FL-TFC with Domain mapped method eliminates the need for deep and complex neural network architectures while ensuring accuracy, efficiency, and strict satisfaction of boundary conditions as compared to standard PINN.

arXiv:2604.24805v1 Announce Type: new Abstract: Modern machine learning optimizes for accuracy without explicitly accounting for internal computational cost, even though physical and biological systems operate under intrinsic energy constraints. We evaluate energy-aware learning across 2,203 experiments spanning vision, text, neuromorphic, and physiological datasets, using 10 seeds per configuration and performing a factorial statistical analysis. Three findings emerge. First, architecture alone explains negligible variance in accuracy (partial eta^2 = 0.001). In contrast, the architecture x dataset interaction is large (partial eta^2 = 0.44, p < 0.001), demonstrating that optimal architecture depends critically on task modality and rejecting the assumption of a universal best architecture. Second, a controlled lambda-sweep over four orders of magnitude validates a single-parameter energy-regularized objective L = L_CE + lambda * E(theta, x): internal activation energy decreases to 6% of baseline at moderate lambda with no accuracy degradation on MNIST. Third, energy-first architectures inspired by an action-principle framework yield 5-33% within-modality training-efficiency gains over conventional baselines. These results emerge from a research program that interprets learning through a structural correspondence between the action functional in classical mechanics, free energy in statistical physics, and KL-regularized objectives in variational inference. We frame this correspondence as a design hypothesis rather than a derivation.