How unit economics should set your classification cutoff, and why they rarely do. The post Your Churn Threshold Is a Pricing Decision appeared first on Towards Data Science.

Tim Bradshaw / Financial Times: Q&A with Luciana Lixandru, who co-leads Sequoia's global early-stage investment business, on why it is time for “act two” for Europe's tech sector, AI, and more  —  The global co-lead of the US venture capital firm's early-stage investment business talks about why it is time for ‘act two’ for Europe's tech sector

The corporate venture arm ends after 10 years and 80 investments.

Hackers claim they stole 1.3TB of Novo Nordisk data, including clinical trial and AI model information, after issuing a $25 million demand. The post Ozempic Maker Novo Nordisk Confirms Security Incident After $25M Hacker Demand appeared first on TechRepublic.

Google's new smart speaker is more about Gemini than audio quality.

Sharomka/Shutterstock Children are innately curious, and throughout any given day they come up with all manner of questions: Why don’t fish have hair? Why do flowers wilt so quickly? Their need to understand the world – and develop their language skills and ideas – makes them tireless conversationalists. While their inquiries would usually be directed at parents or teachers, in modern homes even the youngest kids might now talk to a digital interface like Siri or Alexa. These AI systems are fast becoming part of many children’s everyday lives, as kids ask them to play music, help with their homework, answer questions, or just chat to them. These kinds of interactions are no longer strange, but we need to ask what happens when they become completely routine. Do they change the way children learn to communicate? Do they change the words they use? And are they a threat to kids’ cognitive abilities? Language learning Learning to speak has never been a question of just learning words. Children acquire language through human relationships, and by building emotional ties to other people. They learn to take turns, how to interpret silence and context, and how to tell when someone is tired, annoyed or distracted. They also discover that conversations do not need to be perfect – there will always be interruptions, misunderstandings, and off-the-cuff explanations. But AI does not think like a human. Think about your interactions with ChatGPT or Gemini. We rarely lose our patience while talking to these virtual assistants, partly because these interactions are, by their very nature, governed by a very different logic to human conversation. These tools are built for quick responses and infinite patience, and this changes the experience of communication. AI and politeness In many homes, something very curious is becoming increasingly common: some children (and even adults) are adapting their speech so that virtual assistants will understand them better. They speak in simple sentences, and give direct instructions: “play cartoons, open YouTube, tell me a joke”. This kind of speech – known as instrumental language – aims to get immediate results. This shift does not necessarily mean children are becoming ruder or less empathetic, but it may influence their expectations of conversation in general. Human interactions are usually slow and ambiguous, and require patience, attention and negotiation. Chatbots, on the other hand, are designed to give quick, fluid responses, or even create a sense of virtual empathy with the user. ChatGPT’s artificial empathy is a language trick. Here’s how it works This all leads us to a question that may seem minor, but reveals a lot: should we teach children to say “please” and “thank you” to Alexa? Beyond the surface-level question of whether we should be polite to machines, this debate forces us to think about the communicative habits that children develop through daily interactions with technology that always obeys them. The wider question for families and educators alike is: what idea of “conversing” will children construct in this context? Alongside these doubts, we should not lose sight of the opportunities that these systems present. Many children feel freer to ask questions when they do not fear judgement, and a chatbot will repeat an explanation as many times as is necessary, adjust the level of complexity, or support them as they learn new languages or concepts. These tools provide a safe space for trial and error, free from the social pressures that often accompany human conversation. This is not just the case for children. Many of us now resort to AI to ask ordinary questions, from “Alexa, how do I recover my password?” to more embarrassing queries that we would rather not voice out loud. Chatbot teddies for three-year-olds? Why AI toys are risky for kids Responding is not understanding Current AI systems produce extremely convincing answers, but they do not understand the world the way a person does. They do not have experiences, emotions, or intentions – even if they talk like they do. Just like many adults, young children tend to attribute human qualities to the things they interact with. If something can converse, it is easy to presume that it also has understanding or knowledge. However, a lot of the information in human conversation is unspoken. An adult can tell when a child’s question is the product of curiosity, fear, or a simple need for attention. This pragmatic dimension – consisting of gestures, tone, looks, feelings – is crucial for children’s development. It is difficult to replicate in a machine, which can only offer an answer without capturing any of this nuance. Seven tips for talking to children and young people about generative AI Humans are not machines When children grow up surrounded by a particular kind of linguistic exchange – one that consists of quick responses and having every single request obeyed – it ends up shaping habits, expectations and ways of interacting. This can lead children to always expect clear, quick, effortless answers, as though any conversation were something to be resolved on the spot. The adults who live with young children therefore have a vital role to play. They are the ones who mediate daily use of these tools both at home and at school, who understand their limitations, and who are able to integrate these conversations into general learning. A child asking Alexa to answer their questions or tell them a joke is not, in and of itself, detrimental to their language development. But we should guide these conversations so that they understand they are dealing with a machine that responds to them, not a person. We need to show children what separates us from machines, how we should interact with them, and in what situations it is alright to use them. We should accompany them in these everyday interactions, commenting on them and helping them to understand AI’s limitations. AI can be useful as a support, but under no circumstances should it take the place of replace conversation between people. Despite rapid developments in technology, human interaction remains at the heart of the way we exist in the world. A weekly e-mail in English featuring expertise from scholars and researchers. It provides an introduction to the diversity of research coming out of the continent and considers some of the key issues facing European countries. Get the newsletter! Clara Macarena Ponce Romero is part of the Koiné group at the University of Santiago de Compostela. She currently participates in the project entitled "Agencia Estatal de Investigación, Corpus y densidad de datos. Versión robusta del ‘corpus Koiné’ de habla infantil" (PID2024-158897NB-100), financed by the European Regional Development Fund and the Spanish Ministry of Science, Innovation and Universities.

"Space Launch Complex-6 represents six decades of American innovation."

Muvija M / Reuters: The UK CMA orders Google to provide greater transparency on how its search rankings work and to let users transfer their search data to authorized third parties  —  Britain's competition watchdog ordered Google to provide greater transparency on how its search rankings work …

A Microsoft and Huazhong University benchmark tested GPT-4o, GPT-5, Grok-3, and others on realistic enterprise data scenarios. Privacy violation rates hit 50.9%. More capable models made it worse, and the fix has nothing to do with model selection...

Yohan Yunstaff / Cointelegraph: Trace Finance, which provides stablecoin settlement infrastructure for cross-border payments across Latin America, raised a $32M Series A led by CoinFund  —  Cointelegraph is committed to independent, transparent journalism.  This news article is produced in accordance with Cointelegraph's Editorial Policy …

“People just really want to take back control of their time, their lives, their attention... They’re down for whatever helps them do that.”

Gene therapy company UniQure had another FDA meeting after Vinay Prasad's exit.

If physical AI is going to match the accomplishments of LLMs, there's a data problem that needs to be solved.

Why production-level AI optimization modeling agent needs reproducibility and portability, and how IR helps achieve them The post The Secret to Reproducible and Portable Optimization: ORPilot’s Intermediate Representation (IR) appeared first on Towards Data Science.

Jennifer Smith is listing this home, nicknamed Kentucky Cottage, on Airbnb during the FIFA World Cup. Zerb Mellish for BI The FIFA World Cup will have matches in Arlington, Texas, a suburb of the Dallas-Fort Worth metroplex. It's good news for local Airbnb hosts like Jennifer Smith, who has seen demand surge. Smith told Business Insider her two homes in McKinney, Texas, are nearly booked for June and July. The countdown to the FIFA World Cup is on, and people and businesses across the country are looking for ways to cash in on one of the world's biggest sporting events. Millions of fans are expected to travel to host cities, fueling demand for everything from restaurants and entertainment to places to stay. Airbnb is betting on it. The company is offering $750 to first-time hosts in any of the 16 World Cup host cities across the US, Canada, and Mexico who rent out their homes by July 31, 2026. In a May press release, Airbnb said that since it began its outreach in 2025, more than 100,000 new short-term rentals have been listed in those cities. In Dallas, one Airbnb host is charging $9,130 for a one-night stay in a modest one-bedroom, one-bathroom rental on June 22nd — the same day Argentina and Austria are scheduled to play at the AT&T Stadium in Arlington. Airbnb said 77% of entire-home listings still available for tournament stays are priced below $500 a night. This as-told-to essay is based on a conversation with Jennifer Smith, 55, who is renting out her McKinney, Texas, cottages on Airbnb during the FIFA World Cup. The conversation has been edited for length and clarity. The Kentucky Cottage is located in McKinney, a suburb of Dallas. Zerb Mellish for BI There are a lot of events happening in Dallas this summer, and the World Cup is definitely one of the biggest. There's been so much buzz around the city, everyone's excited. I'm not from a soccer family, so I didn't realize just how big it was until my husband and I started getting Airbnb bookings from people coming from out of the country. That's when I thought, "Hey, this is a big deal." We own two Airbnb properties just north of McKinney, a suburb of Dallas. They're about 45 minutes from the AT&T Stadium in Arlington, where some of the World Cup matches will be held. Smith designed Kentucky Cottage to feel homely and welcoming. Zerb Mellish for BI There's a lot of demand right now, and prices have adjusted to reflect that. As a host, that's been really encouraging to see. It's also been exciting getting our rentals ready for the people who are coming. I've been researching ideas and even using ChatGPT to think through ways we can welcome our international guests and introduce them to Texas and Southern hospitality. Our Airbnbs are booked and busy My husband and I have done long-term rentals with other properties we've owned, but short-term rentals are new to us. Our Airbnb rentals are two next-door cottages we've named the Kentucky Cottage and Derby. They're both two-bedroom, one-bathroom homes. Between the two of them, I'd say we spent about $75,000 renovating them. That's definitely on the high end, but these are old, historic homes that were not in good shape. Smith wants the guests to feel as comfortable as possible during their stay. Zerb Mellish for BI On a slow week, our normal rate for each cottage is about $155 per night. On weekends, it usually jumps to around $195 or $200, and during holidays, it's typically closer to $250 a night. For pricing stays during the tournament, I use Airbnb's Smart Pricing tool. Basically, I give them my lowest and highest price, which is much more than I would expect to get normally. A bedroom in Kentucky Cottage and its kitchen. Courtesy of Airbnb For June and July, our rate has about doubled compared with a typical night. I've looked at what hotels in the area are charging, and our pricing feels in line with the market. Because of all of the demand, we're definitely doing a lot better this year than we did last year. We're fully booked for June and have some availability in July. I wouldn't have thought that many people would want to come to Texas during the summer because it's pretty hot here, but the World Cup and other events are bringing them. We want to make our guests' stay special Smith prepared this welcome basket for her guests visiting from Japan. Zerb Mellish for BI Our World Cup guests have booked stays anywhere from two to seven days. They're coming to see their team. But outside the stadium, what really matters is how welcome and connected they feel. My husband and I are working with some local vendors on welcome baskets. We're hoping to include a few things that are specific to where our guests are coming from, along with some Texas and American touches — maybe even bourbon drinks to celebrate this year's July 4th, America's 250th birthday. We've also ordered flags for the different countries, and both cottages have welcome boards. When guests arrive, we'll have something written on them, like "Welcome" and "Best of luck to your team." We want them to feel like we're excited they're here and that we're cheering them on, too. The Kentucky Cottage is nearly booked for the entire month of June and nearly all of July on Airbnb. Zerb Mellish for BI I'll need to be fully aware of where guests are coming from, which team they're supporting, and even take the extra step to see how their team is doing. And with world events, I'll also need to be mindful of what's happening in the news. It's a lot of work, but worth it Our hope is that everything we're doing, and the money we earn from the cottages, can eventually roll into another property. We always have our feelers out, looking for similar properties — older homes that need work but have potential. The living room of Kentucky Cottage. Zerb Mellish for BI I know a lot of people are hesitant to list their homes on Airbnb, and I was, too. I wondered whether it would be safe to have people in my home. How would they take care of it? But it's been a fabulous experience. We haven't had any issues with our house or the neighbors. I'd encourage anyone thinking about hosting during a major event like the World Cup to do it. It takes time and effort, but it's a rare chance to extend a hand to the world that's coming here — I think our world can use that right now. Kentucky Cottage's backyard. Courtesy of Airbnb We love our community, and we want people to experience it. This feels like such a great opportunity to get to know people, be a blessing to them, and be blessed in return. Read the original article on Business Insider

Discussion | Link

Anthropic CEO Dario Amodei is attending a CEO lunch at G7 along with OpenAI CEO Sam Altman and Google Deepmind's Demis Hassabis on Wednesday. Anna Moneymaker/Getty Images A lunch today in France will see the leaders of the fiercest AI rivalry break bread. Dario Amodei and Sam Altman will discuss how to keep societies resilient, especially for young people, POLITICO reports. Anthropic's ongoing dispute with the White House over Fable 5 and Mythos 5 will be the "elephant in the room." European leaders are meeting on Wednesday with the US president and Anthropic CEO Dario Amodei just days after Washington suspended access to Anthropic's latest AI models. Don't expect them to confront Donald Trump. A lunch with the world's leading artificial intelligence CEOs at the G7 meeting in Evian-les-Bains offers the first opportunity for EU leaders to ensure their voices are heard after the US blocked EU citizens from accessing Anthropic's cutting-edge tech — a move that triggered renewed calls for Europe to become less reliant on US artificial intelligence. But there are no signs Europe is on the warpath. Despite the hostile move from Washington, diplomats and officials from nations attending the gathering insist they can work with the US on minimizing AI's security risks — seeking to turn the episode into a launchpad for collaboration rather than a spat that drives the continents apart. "We are ready to engage and tackle these security risks together with our like-minded partners," European Commission spokesperson Thomas Regnier told reporters in Brussels on Tuesday in advance of the G7 get-together. "On the topic of frontier models, we should be able to create unity before the end of the G7," a European diplomat said. "The question is to recreate confidence, we need to recreate a circle of trust." Wednesday's lunch is — on paper at least — not meant to address the threats AI poses to cybersecurity, or the dramatic events of the past days that illuminated inconsistencies in Washington's approach to the most advanced AI. During a two-and-a-half-hour lunch, CEOs including Anthropic's Amodei, OpenAI's Sam Altman, DeepMind's Demis Hassabis and Mistral's Arthur Mensch will discuss how AI can drive economic growth and how to keep societies resilient, especially for young people, according to the official agenda. Yet the spat between the Trump administration and Anthropic will be the "elephant in the room," according to one industry representative, who declined to be identified because they couldn't comment on preparations for the meeting. Anthropic confirmed that Amodei would attend the meeting, but declined to comment further on the G7 discussions. Confirming the company would meet Commission and EU cyber authorities in San Francisco on Thursday, an Anthropic spokesperson said: "The visit is part of our ongoing engagement with the EU, allied democracies and important international institutions on frontier AI's implications for cybersecurity and opportunities for international collaboration." Under the bloc's AI law, providers of frontier AI models already face strict obligations to test and evaluate their models for a series of risks. Brussels is navigating the matter cautiously, as there has been no formal communication or notification from the US government regarding export controls — only the statement that Anthropic issued on Friday. In response to the government slapping restrictions on the company's new models to bar non-US citizens from using them, Anthropic said it had cut off global access to Fable 5 and Mythos 5, its newest, most tightly controlled models with elite cyber capabilities. The EU's tech chief, Henna Virkkunen, told European Parliament lawmakers on Tuesday that "contingency measures taken in this light should not be discriminatory against partners," referring to the US order against Anthropic. But a read-out of a bilateral meeting between Commission President Ursula von der Leyen and Trump at the G7 on Tuesday didn't mention the issue. "There will be discussions at G7 with tech companies, I will not be prejudging now what will be raised or not in that context," said Arianna Podestà, the European Commission's deputy chief spokesperson. Italian Prime Minister Giorgia Meloni will use the lunch to defend a position Rome has been pushing since it chaired the G7 back in 2024, an Italian diplomat said. That focuses on the "responsibility" for AI and on the definition of truth in artificial intelligence, they said — amid a debate over watermarking AI-generated content, which Meloni also discussed Monday with Japanese Prime Minister Sanae Takaichi in Rome. "It is more and more difficult to make the difference between what is true and what is fake. This is definitely a topic that [Meloni] will raise," the diplomat said. One of the few European tech executives set to participate in the meeting, Domyn's Uljan Sharka, told POLITICO that Europe needs to find common ground with the US. "This narrative of us versus them is completely wrong, and I hope that during the G7 this is going to be addressed," Sharka said. He added that the transatlantic partnership, as it exists for defense with NATO, "should also work on AI." That didn't stop him from picking sides in the fight between Anthropic and the White House. "I don't blame the US administration for doing that," he said. "They were pushed and forced to take action," referring to Anthropic's branding of its models as highly capable of finding software vulnerabilities. The U.K. government said it had been in touch with the US government and Anthropic to understand the situation. Following the export control order, the U.K.'s AI Minister Kanishka Narayan said "the main lesson" was that "as we debate the future of national security and technological sovereignty, access to AI capabilities is crucial." Clea Caulcutt, Giorgio Leali and Joseph Bambridge contributed to this report. This story originally appeared on POLITICO and is courtesy of the Axel Springer Global Reporters Network, which harnesses the resources of the company's newsrooms to publish ambitious scoops, investigations, interviews, opinion pieces, and analysis. It allows journalists — including those from POLITICO, Business Insider, WELT, BILD, Onet, and Fakt — to collaborate on major stories for an international audience of hundreds of millions across platforms. Read the original article on Business Insider

Toward Generalist Autonomous Research via Hypothesis-Tree Refinement

Pramaana will focus on highly sensitive verticals like law, drug discovery, and tax preparation — where errors can be costly and reliability is at a premium.