Feeling the winter chill? Thermal baths and scenic outdoor saunas worth visiting in Europe
Whether it’s floating saunas in Norway or steaming thermal baths in Budapest, Europe offers travellers some spectacular spots to take on the winter chill.
AI/ML news, top picks, and generated innovation digests.
88316 matching items
Whether it’s floating saunas in Norway or steaming thermal baths in Budapest, Europe offers travellers some spectacular spots to take on the winter chill.
Catch up with the most important stories from around Europe and beyond this November 15th, 2026 - latest news, breaking news, World, Business, Entertainment, Politics, Culture, Travel.
Jensen Huang’s affair with a suspected spy is Silicon Valley’s worst nightmare come true The resignation this week of Nvidia’s founder and CEO Jensen Huang is one of the most astonishing corporate scandals in memory, featuring allegations of international espionage and extramarital affairs involving one of the world’s most powerful men. As investigations mount into what went wrong at the $4tn chip company, the saga may also become a milestone in the decades-long history of spies infiltrating the US tech industry. Huang admitted to being unfaithful to his wife of 41 years in a letter to employees this past weekend, following a report from the New York Times that the tech mogul had a close relationship with a woman US officials suspect of spying for China’s government. Huang resigned on Thursday after 33 years as CEO amid reports of two additional affairs with Nvidia employees, causing the world’s most valuable chip company’s stock to plummet. Continue reading...
Epistemic status: Low, but I hope to be wrong in interesting ways As a non-computer scientist-scientist [1] I find it hard to think about the varying perspectives on Artificial Intelligence alignment difficulty. To make this more intuitive I invoke the 3D surface of a fitness landscape like structure as a visual representation. [2] The AI Safety Landscape AI safety is thought to be largely pre-paradigmatic, although of the current views AI safety perspectives there are noticeable camps, and within each there is wide variation. It is also rapidly evolving as more information comes out about their behavior that allows us to update. Previous work categorizing these has focused on specific technical approaches as well as their feasibility . Here I think of ‘alignment’ loosely, here it is represented as any mind that would not cause extinction (or an s-risk) while also having substantial power, this contrasts with misalignment which would lead to extinction (or an s-risk). To simplify this I condense these to five main camps represented by their arguably most prominent opinion havers: Although I agree that P(doom) is a Dumb Meme it is used here more as a heuristic for parsing levels of concern around AI. For more of their specific positions here are my references: Yann LeCun , Paul Christiano [3] , Eliezer Yudkowsky [4] , Daniel Kokotajlo [5] and Roman Yampolskiy . Visual Representation In my mind’s eye, the alignment landscape manifests similarly to a fitness landscape . [6] A f…
Anthropic's IPO roadshow video has a graph. It shows four models; I think dot 1 = Claude 3.5, dot 2 = Claude 4, dot 3 = Mythos, arrowhead = unreleased model. The graph provided doesn't have any numbers on the Y axis. In this post I'll explain why I think the Y axis lines most likely represent a 10x jump each, and so the point at the end of the graph represents about a 30000x CM (i.e. Anthropic can currently pretrain models at the same quality with 1/30000th of the compute it needed in 2024). I'm pretty unsure about this though and it could plausibly be much higher or much lower. Keep in mind that this graph is only about pretraining and doesn't say anything about post-training. [1] Without extra context, it's hard to say how much the compute has actually been multiplied: if each Y gridline is 1.1x the total multiplier would be about 1.53x; if each Y gridline is 10x the total multiplier would be about 30000x. (It's not clear to me why the last point is an arrowhead instead of a dot - maybe it's supposed to be their guess of what the compute multiplier will end up at for their latest pretraining run?) What people in the IPO roadshow video say about compute multipliers Sam McCandlish said (while that graph was displayed): It’s how we serve models to customers and how we use Claude at scale internally. By training what we believe to be the most efficient models, we can do more of all these things with the same amount of compute. It’s also how we drive down costs, both for custom…
Plus: Cheap solar, who pays for Claude & DNA immunotherapy++
In January 2024, Ryan Greenblatt and Buck Shlegeris posted " Catching AIs red-handed ." This post got on "Best of LessWrong" in 2025, and in December of last year Buck Shlegeris commented on it explaining that he was disappointed it did not lead to more follow-up research. I upvoted the post at the time of the review, and so it seems like it's representative of the analysis the community values. As it happens, it turned out to be remarkably prescient and near-term-relevant for a theorypost: just this August 2026 we, in fact, got to see what happens when AI companies "catch their AIs red-handed". The parties who wrote the post were, additionally, some of the very people involved in investigating the Hugging Face incident. I definitely respect these researchers, but I struggle to imagine how this analysis could have been useful. While it seems possible that publishing research like this is what put Redwood in a position to be called upon to investigate in the first place, the piece is just a long and abstract discussion about how (for example) if you catch an AI scheming, it is less likely to succeed (emphasis not mine): In this post, we’ll discuss a crucial dynamic related to risk from scheming models. Suppose your model is trying to exfiltrate its weights or hack the datacenter or cause some other difficult-to-reverse bad outcome. (We’ll refer to all of these as the model “escaping”, because escape is linguistically convenient and is a good description of many of these diffi…
Release: datasette-apps 0.4a0 Datasette Apps are sandboxed HTML and JavaScript applications that live inside Datasette - think Claude Artifacts, but these ones can run SQL queries against your data. 0.4a0 adds support for linking: you can link to an app with #fragments in the URL and the app can see and use those, plus supported URL schemes are reported back to Datasette Agent so it can help write apps that link to each other. This means you can have e.g. an app that shows details about a specific record, and a completely separate app that searches those records and links to their representations in the first app. Tags: datasette-agent , datasette-apps , datasette , urls
구글이 \'제미나이\'에 구글 월렛(Google Wallet)을 연동해 개인 금융 정보를 분석하는 기능을 추가했다. 사용자의 소비 내역과 금융 계좌 정보를 바탕으로 지출 패턴을 분석하고, 예산 관리와 저축 계획까지 제안하는 기능이다.구글은 13일(현지시간) 공식 블로그를 통해 제미나이의 개인화 기능인 \'퍼스널 인텔리전스(Personal Intelligence)\'에 구글 월렛을 새롭게 연결한다고 밝혔다. 미국 내 모든 제미나이 앱 사용자를 대상으로 웹과 안드로이드, iOS에서 차례로 제공한다.사용자는 설정의 \'연결된 앱(Connected
We’re going to see huge, foundational cybersecurity attacks given the rapid rise in AI capabilities. Having worked in hostile environments, I’ve thought a lot about physical and digital security over the years. While you may not be personally targeted, or doing dangerous work, your data is undoubtedly leaked somewhere, and the digital providers you use—for email, banking, social media, etc.—are more and more likely to get hacked. Discuss
There was an AI protest day today, with parallel protests around the world, and our family went to the Boston one: I'm not really a protest person : normally I share my views in writing . But while there's been a lot of writing on the internet about the risk that AI gets us all killed (some of it from me ), this can't be an internet-only thing. Way too few people are taking seriously the risk that OpenAI, Anthropic, and others are rapidly building something that could actually cause human extinction. A protest is a literal demonstration that people care, and it expands visibility to the general public. We talked with the kids about whether they wanted to come, or whether I should take them somewhere while Julia went to the protest. All three did want to come, which makes sense: they've been hearing us talk about this at the dinner table and with friends for months, and the risk impacts them at least as much as it does us. We painted some signs: Clementine was not helpful: Mine was a bit on the long side: The crowd was mostly students, but there were also a lot of people I recognized from Boston EA meetups . Favorite sign: an "If You Can't Steer Don't Race", painted by someone I'd run into at LessOnline . It was nominally a protest demanding a US–China treaty that bans more powerful AI until it can be done safely, but in practice it was a much more high level protest around the "slow down AI", "regulate AI", "AI bad", "RSI will get us killed", "AI companies irresponsible" clu…
Release: datasette-agent 0.6a0 Accumulated small fixes for Datasette Agent, my agent plugin for Datasette that lets you ask questions about your data. Most notably, it now adds an agent prompt directly to the Datasette homepage, with support for hash-mentions of tables and queries: Tags: datasette-agent , llm , datasette , llms , ai , generative-ai
테슬라가 일부 고객을 대상으로 완전 자율주행차 \'사이버캡(Cybercab)\'의 개인 구매를 위한 사전 주문을 시작했다. 차량 가격은 약 4000만원이지만, 운행하려면 매달 40만원의 구독료를 내야 하고 이용 지역과 상업적 운행도 제한된다.테슬라는 13일(현지시간) X를 통해 일부 고객에게 개인용 사이버캡 사전 주문 기회를 제공한다고 발표했다. 이는 로보택시 전용으로 개발한 자율주행차를 일반 고객에게 직접 판매하려는 첫 단계다.공개된 주문 페이지에는 차량 가격 2만9990달러와 월 299달러의 \'오토노미 구독(Autonomy Subs
메타가 직원들을 대상으로 예측 시장 앱 \'테이크(Takes)\'를 테스트하고 있다. 실제 돈을 걸지 않고 가상 포인트로 각종 사건이나 이벤트의 결과를 예측하는 방식으로, 칼시나 폴리마켓과 달리 현금 거래를 지원하지 않는 것이 특징이다.월스트리트저널(WSJ)은 13일(현지시간) 메타 직원들이 개발 중인 예측 시장 앱을 시험하고 있다고 보도했다.이용자는 \'크레드(Cred)\'라는 가상 포인트를 걸고 특정 사건의 결과를 예측한다. 예상이 맞으면 포인트를 얻고 틀리면 잃는 구조다.테스트에 등장한 질문에는 팀 쿡 전 애플 CEO가 차기 엔비디아
중국에서 엔비디아의 수출 제한 AI 칩을 기반으로 추진 중인 4조원대 데이터센터 사업이 차질을 빚을 가능성이 커지고 있다. 미국의 규제 강화 움직임과 젠슨 황 전 엔비디아 CEO의 사임으로 핵심 GPU 확보가 불투명해진 가운데, 중국 기업들은 여전히 엔비디아 칩을 전제로 대규모 투자를 진행하고 있는 것으로 나타났다.로이터는 13일(현지시간) 중국의 조달 공고와 투자 계획, 건설 계약, 기업 공시 등 150건 이상을 분석한 결과, 엔비디아의 수출 제한 프로세서를 기반으로 한 프로젝트와 장비 계약 규모가 230억위안(약 4조2000억원
What is the oldest text in Claude’s training data? Claude was trained on a non-trivial fraction of everything humans have ever written down. That includes the things you’d expect, like books and research papers. But it also includes microfilm archives, transcribed Betamax tapes, Donald Trump’s prenups, ancient scrolls and Harry Potter fanfiction. It’s a weird library. So what is the oldest text in Claude's training data? It mostly depends on what we mean by "text". Here are some candidates: The Epic of Gilgamesh The Epic of Gilgamesh is considered the world's oldest written "great work of literature"; the earliest versions of it were written in 2100 BC in Sumer. The Sumerians did a fair amount of writing. Their medium was clay tablets. These tablets were designed to endure a few seasons of the Mesopotamian weather; this turned out to generalize well, and instead of surviving for a few years, they survived for millennia. Gilgamesh was lost in the sands of Iraq for a while, but was eventually dug up by European archeologists. The scholar who first translated the Epic to English got so excited by it that he took all of his clothes off and ran around the English museum shouting "I am the first man to read that after more than two thousand years of oblivion." tablets being scanned In the 1980s and 90s, scholars digitized many of the dug-up clay tablets. The museums would scan the tablets in normal office scanners and fax them to Assyriologists to phonetically transliterate. Those…
Ukraine is countering Russia's narrative of battlefield success by blunting the Kremlin's attacks.
이탈리아 개인정보보호 당국이 안드로이드 사용자의 동의 없이 스마트폰 사진과 동영상에 접근한 메타에 8억8700만달러(약 1조1886억원)의 벌금을 부과했다. 페이스북 앱의 사진 추천 기능이 사용자에게 충분히 고지되지 않은 채 개인 갤러리 데이터에 접근했다는 이유에서다.이탈리아 당국은 14일(현지시간) 발표를 통해 메타가 페이스북 앱에서 사용자의 사진과 동영상을 확인하고, 게시할 만한 콘텐츠를 추천하는 기능을 운영하면서 별도의 명확한 동의를 받지 않았다고 밝혔다. 이탈리아에서 약 2000만명의 안드로이드 사용자가 영향을 받은 것으로
젠슨 황 전 엔비디아 CEO의 불륜과 사임 논란이 생성 AI를 활용한 풍자 영상과 이미지로 이어지고 있다. 영화 \'매트릭스\'의 주인공부터 뮤직비디오 속 가수까지 황의 얼굴을 합성한 영상들이 수백만 조회수를 기록하고 있다. AI 산업을 성장시킨 황이 자신이 공급해 온 GPU와 AI 기술로 만들어진 밈의 주인공이 됐다는 점이 눈길을 끈다.최근 X와 인스타그램, 레딧 등에는 황을 풍자한 AI 생성 영상과 이미지가 잇따라 등장하고 있다. 영화 속 인물로 바꾸거나 불륜과 중국 스파이 의혹을 패러디한 콘텐츠가 주를 이루며, 일부는 수백만회 이
SpaceX’s Starship returns from orbit and is captured by launch pad arms in a dramatic first
스페이스X가 초대형 우주선 \'스타십(Starship)\'을 지구 궤도에서 귀환시켜 발사탑의 기계식 팔로 붙잡는 데 처음으로 성공했다. 약 22시간 동안 지구를 14바퀴 돈 우주선을 회수하면서 완전 재사용 로켓 개발에 중요한 진전을 이뤘다. 그러나 1단 로켓 \'슈퍼헤비\'는 귀환 과정에서 문제가 발생해 파괴됐다. 목표였던 부스터와 우주선의 동시 회수에는 실패한 것이다.스페이스X는 14일(현지시간) 15번째 스타십 시험비행에서 상단 우주선의 첫 발사탑 포획에 성공했다고 밝혔다. 전날 텍사스주 스타베이스에서 발사된 스타십은 지구 궤도에 진입
We are in AA 9. In Brave New World, Aldous Huxley dated the future from the launch of a car. The book was set in AF 632; AF stands for "Anno Fordi" or "After Ford". It starts counting from 1908 AD, the year when Ford launched the Model T into production. In the book, people say "Ford!", instead of "God!", and the cross has been cut down into a T. It's meant to be humorous. But the point here is that a society built around mass production might end up dating itself by it. The case is not so different for us; instead of mass production, we have AI, and we haven't picked our 1908 yet. There are a few obvious candidates, such as Turing's 1950 paper, which asked the gold question: can machines think? The 1956 Dartmouth workshop , which coined the term "Artificial Intelligence". We then have ChatGPT in November 2022, which accelerated a lot, and ChatGPT is the Model T of the AI era, in my opinion. But AI is different in the sense that it is a collaborative result of the efforts of a lot of people, research institutions, and the compute industry, among others. It might not be wise for us to go by the Model T analogy and choose a person or a product. And we have a better alternative. My pick is June 2017 , when a team of eight researchers, mostly from Google, published the paper , " Attention Is All You Need ", and it had huge downstream effects. And dating our era from a paper instead of a person or a product kind of feels right. It was an idea, written down by a team, published op…
Following my 3-week visit to Berkeley as described here and featured by ACX here , I told various people I would write in detail why I didn’t stick around to work in AI safety. In this post, I will dance around one object-level disagreement with AI safety and rationalists: why I no longer believe in explosive world transformation due to ASI. There will eventually be follow-up pieces on more object-level arguments and meta-level arguments against AI safety, but I prefer for this post to be mostly self-contained and not explicitly about AI safety. When I worked in theoretical physics research, I was constantly reminded that if I were just that much smarter and had read that many more papers, I would be able to make the right connection to solve the problem. You are always the bottleneck; it’s always a skill issue. So if you’ve spent the bulk of your conscious life honing your ability to manipulate the pure abstraction of math and computer code, it is staggering to watch the problems you wrestled with for months or years get blown away in hours. You are left wondering how quickly all of the other problems in the world will be resolved - all of disease perhaps? Why not all of biology? Why not all the secrets of the universe? In December 2025 when Claude Code + Opus 4.5 first made agents feel extremely useful, I was concerned about the job market being obliterated, so I took out “insurance” on rapid-onset white-collar unemployment . Thankfully, those predictions about increasingl…
Jensen Huang's supporters question espionage allegations. But Nvidia's board may have had other reasons to accept his resignation.
My wife and I are planning to have children soon, and we will be using embryo selection to improve our family’s future well-being. Basically, the idea is that we can take a small sample of cells from each embryo, analyze their DNA, and use the genetic information to help choose which embryos we want to implant. These days, polygenic scores are good at predicting risk for a lot of diseases, like cancer, heart disease, diabetes, schizophrenia, and Alzheimer’s. And companies like Herasight or Genomic Prediction often give an “embryo health score” [1] that takes an average of the predicted risk for each disease, weighted by how bad each disease is. But how are those weights calculated? And can we do better? Disability-adjusted life years Diseases are bad in two ways: they can make you die, and they can make your life worse. The “disability-adjusted life year” (DALY) metric captures both of these, by combining years of life lost and years of reduced quality of life (using a factor between 0 and 1 to account for different subjective severity of disease states). [2] Embryo selection companies, who compute polygenic scores for many different diseases, can add up the DALY burden of all the disease predictions for a given embryo. This gives an overall “embryo health score” that is a composite of these predictions. DALY scores for five of our embryos, shown against the possible distribution of embryos given the parents’ DNA. Note the left-tailed skew. This is because an increasing gene…
목포 고하도가 해양수산부의 2027년 ‘어촌뉴딜3.0 어촌회복형’ 사업 대상지로 선정됐다. 국·도비 79억원을 포함해 2027년부터 2030년까지 총 100억원이 투입된다.목포시는 고하도가 해양수산부 어촌뉴딜3.0 공모사업 대상지로 최종 선정됐다고 12일 밝혔다. 사업비는 국비 70억원과 도비 9억원, 시비 21억원 등 총 100억원이다.목포시는 사업을 통해 충무 폐교 부지에 체류형 문화복지 플랫폼을 조성하고 고하도 마을안길 포장과 마을연결로 개설 등 생활 기반을 개선할 계획이다.태양광·에너지저장장치(ESS) 설치와 제방 보강, 선
루마니아에서 전력망 접속 기술승인을 받은 재생에너지 발전사업 용량이 114.9기가와트(GW)에 달하지만, 접속계약과 건축허가, 발전시설 설립허가까지 확보한 사업은 16.2GW에 그친 것으로 집계됐다. 전체 접속승인 용량은 전월 대비 소폭 감소했으나 주요 인허가를 모두 확보한 사업 규모는 5.28% 증가했다.루마니아 에너지규제청(ANRE)이 지난 9일(현지시간) 발표한 재생에너지 발전사업 현황에 따르면, 10월1일 기준 최대 계통 송출 승인용량이 1메가와트(MW) 이상인 사업 중 유효한 전력망 접속 기술승인(ATR)을 보유한 사업은
This post was written as part of the Iliad Fellowship , where I am mentored by Dmitry Vaintrob and collaborate with Brianna Grado-White. The initial inspiration came from conversation with Richard Ngo and Adrià Garriga-Alonso. To all of these my thanks. I started writing this essay to share some thoughts on how to reconcile the idea of circuits with gradient descent. In doing this, I've come to change my perspective pretty substantially. The first half of this essay engages a Darwinian picture of circuits, but I think the central point - that we need a clearer ontology of computation in neural networks - holds without reference thereto. Before, I was thinking about circuits as 'units of selection' - the kinds of things which engage in some gradient-hacking shenanigans to save themselves from gradient descent. I don't think this picture is unimportant or unhelpful, but I've come to think that it sidesteps the hard part of the problem - that the things the circuit Darwinism picture takes for granted are the confused parts, and I think therefore the more insightful to examine. I have gone from thinking "circuit Darwinism, but we need to formalize how to delineate the circuits" to "things that under circuit Darwinism seem like implementation details are actually the deep source of confusion, and we should treat these as the problem". I'll aim to convince you that this change in perspective is the right direction to head in. I wrote this essay in three parts, plus what became a c…
호주 태양전지 개발업체 선드라이브솔라(SunDrive Solar)가 은 대신 구리를 전극에 적용한 초박형 태양전지로 우주선용 태양광 패널 개발에 나섰다. 2027년 4월까지 우주선에 장착할 패널 시제품을 제작하고, 우주 환경을 모사한 시험을 통해 실제 비행용 제품 개발에 필요한 성능을 검증할 계획이다.선드라이브와 호주 우주선 개발업체 스페이스머신스컴퍼니(SMC)는 지난 9일 이 같은 공동 연구개발 계획을 발표했다. 양사 기술진은 지난 10월 1일 호주 커넬에 있는 선드라이브 시설에서 프로젝트를 시작했으며, 1차 개발은 내년 4월까지
베트남이 2026년 태양광 발전가격 상한을 전년 대비 6.9~16.9% 인상했다. 배터리에너지저장장치(ESS)를 결합한 발전소에는 별도로 높은 상한을 적용하며, 북부 수상태양광의 최고 가격은 킬로와트시(kWh)당 2120.31동(약 110원)으로 정해졌다.베트남 산업무역부는 지난 5일(현지시간) 이 같은 내용을 담은 2026년 태양광 발전가격 체계 결정 제2946호를 시행했다. 가격 상한은 지상형과 수상형 태양광을 구분하고, 북부·중부·남부 3개 지역과 ESS 설치 여부에 따라 각각 다르게 적용된다.ESS를 설치하지 않은 지상형 태
A self-hosted Slack clone where Claude, Codex, Gemini and Grok agents join channels as teammates, with approvals, budgets and MCP tools
Nvidia’s security chief quietly left in August, and now that CEO Jensen Huang has stepped down over his relationships, including one with an executive at a company suspected of funneling chips to China, a former intelligence officer says it’s the kind of leak that runs everyday conversations, which most corporate security teams are poorly equipped to catch.
A from-scratch Vietnamese TTS hits 48 kHz audio, instant voice cloning, and real-time CPU inference with an OpenAI-compatible streaming API.