AINews Portal

News

50 items (filtered)

The raw AI news stream — everything we ingest across RSS, Hacker News, arXiv, GitHub, Bluesky, YouTube and more, newest first.

SourceAgeTitleSummaryStatus
bluesky
bluesky-ai
2moI obviously think it's bad if OpenAI stole data from mathematicians (who were using their LLM) and then put out a paper about how smart the…I obviously think it's bad if OpenAI stole data from mathematicians (who were using their LLM) and then put out a paper about how smar…
↗Not briefed
rss
techmeme
2moIreland's media watchdog investigates X over concerns about age assurance measures and parental controls, the first formal probe under the Online Safety Code (Brian O'Donovan/RTÉ)Brian O'Donovan / RTÉ: Ireland's media watchdog investigates X over concerns about age assurance measures and parental controls, the f…
↗Not briefed
arxiv
arxiv-ai
2moGaussian Approximation for Multivariate Martingale Sums from Uniformly Ergodic Markov ChainsWe develop Gaussian approximation bounds in higher-order Wasserstein distance $W_p$, $p\geq2$, for sums of multivariate martingale dif…
↗Not briefed
arxiv
arxiv-ai
2moUnthrottling the Tanh Jacobian in SAC: A Negative Result on Bang-Bang Control and MetaDriveSoft Actor-Critic (SAC) represents a continuous policy as an unbounded Gaussian that is squashed by tanh. The Jacobian of that map is…
↗Not briefed
rss
techmeme
2moBlock says it has submitted an application to US regulators to establish a federally regulated, uninsured national trust bank called Builders Bank & Trust (Elias Schisgall/Wall Street Journal)Elias Schisgall / Wall Street Journal: Block says it has submitted an application to US regulators to establish a federally regulated,…
↗Not briefed
hn
hn-ai
2moShow HN: Routi Bot – AI bots with their own desktops on your Macbeen working on Routi Bot. an open source mac app inspired by Grok Bot.<p>each bot has its own instructions, desktop and can be used w…
↗Not briefed
arxiv
arxiv-ai
2moLeCor: Learning to Be Corrected by Meta-Learned Test-Time Training for Interactive 3D Lung-Tumour SegmentationDelineating lung tumours on computed tomography (CT) takes a considerable share of the time spent on radiotherapy planning, and a cont…
↗Not briefed
bluesky
bluesky-ai
2moOpenAI set $22M on fire just to steal $1M from a human who did the actual work, and that’s the business plan our economy is betting everyth…OpenAI set $22M on fire just to steal $1M from a human who did the actual work, and that’s the business plan our economy is betting ev…
↗Not briefed
bluesky
bluesky-ai
2moEita! 1. Resolveram Navier–Stokes, problema famoso de dinâmica de fluidos 2. A OpenAI afirma que chegou primeiro, mas dois matemáticos diz…Eita! 1. Resolveram Navier–Stokes, problema famoso de dinâmica de fluidos 2. A OpenAI afirma que chegou primeiro, mas dois matemático…
↗Not briefed
bluesky
bluesky-ai
2moOpenAI claims to have solved maths problem that stumped humans for decadesOpenAI claims to have solved maths problem that stumped humans for decades
↗Not briefed
hn
hn-ai
2moAsk HN: Why not change the name to Anthropic News?This site seems like it’s only a mouthpiece for bullshit like Claude and AT Proto - why not just change the name???
↗Not briefed
arxiv
arxiv-ai
2moFrom Fixed Keys to Readable Schemas: Small Language Models for Vehicle Agent Function CallsIn-vehicle assistants must translate natural-language requests into accurate vehicle function calls under strict memory and latency co…
↗Not briefed
arxiv
arxiv-ai
2moDistributed Physical Layer Authentication and Collaborative RSMA in Non-Terrestrial Networks via Graph Reinforcement LearningExisting physical-layer authentication (PLA) schemes for non-terrestrial networks (NTNs) often rely on single-anchor verification, lac…
↗Not briefed
hn
hn-ai
2moThe universal programming language of LLMsHow do you imagine the universal programming language of future LLMs will look? I mean a language built by LLMs for LLMs, not for huma…
↗Not briefed
arxiv
arxiv-ai
2moMode Coverage in Normalizing Flow Boltzmann Generators via Log-Ratio VariationNormalizing flow Boltzmann generators retain a tractable pushforward density, but training with forward KL depends on target samples t…
↗Not briefed
bluesky
bluesky-ai
2moTerrence Tao is close to like the most? i want to say influential person in the field who was very favorable to usefulness of this tool to …Terrence Tao is close to like the most? i want to say influential person in the field who was very favorable to usefulness of this too…
↗Not briefed
bluesky
bluesky-ai
2moyou're telling me that openai used stolen research to solve a millennium prize problem? wowyou're telling me that openai used stolen research to solve a millennium prize problem? wow
↗Not briefed
bluesky
bluesky-ai
2mocto of bluesky describing the backlash against ai chatbots as "futureshock" and then going "i'll never forget the first time i booted up cl…cto of bluesky describing the backlash against ai chatbots as "futureshock" and then going "i'll never forget the first time i booted…
↗Not briefed
rss
techmeme
2moChime agrees to acquire longtime banking partner Stride Bank for $590M; Stride will become Chime Bank, a wholly owned subsidiary; CHYM jumps 8.5%+ after hours (Paige Smith/Bloomberg)Paige Smith / Bloomberg: Chime agrees to acquire longtime banking partner Stride Bank for $590M; Stride will become Chime Bank, a whol…
↗Not briefed
bluesky
bluesky-ai
2moMore broadly, this result is both exhausting and disheartening. I don't like that OpenAI rushed to find a solution and spent 10s of million…More broadly, this result is both exhausting and disheartening. I don't like that OpenAI rushed to find a solution and spent 10s of mi…
↗Not briefed
bluesky
bluesky-ai
2moI’ve never delved that deeply, so I’d need to read their claims. I’d certainly believe this could be a flow where continuity is broken. I’m…I’ve never delved that deeply, so I’d need to read their claims. I’d certainly believe this could be a flow where continuity is broken…
↗Not briefed
hn
hn-ai
2moShow HN: Browser LLM FitAuto-detect client hardware capabilities and match against in-browser AI models across WebGPU, WASM, and ONNX Runtime.<p>GH: <a href="…
↗Not briefed
bluesky
bluesky-ai
2moKeep doing that breathing, it helps.Keep doing that breathing, it helps.
↗Not briefed
arxiv
arxiv-ai
2moBuilding the Harness Automatically: Self-Play in Code Distills a Text Harness for Black-Box OptimizationCan an agent learn a numerical search strategy through executable practice and then transfer that strategy as text? We study low-budge…
↗Not briefed
bluesky
bluesky-ai
2moIt occurs to me that all the drama around the Navier-Stokes result is a tremendous stroke of luck for militant AI-antis: rather than acknow…It occurs to me that all the drama around the Navier-Stokes result is a tremendous stroke of luck for militant AI-antis: rather than a…
↗Not briefed
arxiv
arxiv-ai
2moExact-Form Regret for Gradient Descent, Mirror Descent and Follow-the-Regularized-LeaderOnline gradient descent is usually studied through external regret, where the learner competes with fixed alternatives. Recent work sh…
↗Not briefed
bluesky
bluesky-ai
2moIt’s a bunch of random tables. Given sufficient time and 50 million pages of printouts, you can duplicate any LLM output with some percent…It’s a bunch of random tables. Given sufficient time and 50 million pages of printouts, you can duplicate any LLM output with some pe…
↗Not briefed
bluesky
bluesky-ai
2moCoyote vs AnthropicCoyote vs Anthropic
↗Not briefed
bluesky
bluesky-ai
2mo"While unlikely, we cannot rule out that de-identified data derived from their usage of our products helped improve our models⁠." - OpenAI …"While unlikely, we cannot rule out that de-identified data derived from their usage of our products helped improve our models⁠." - Op…
↗Not briefed
hn
hn-ai
2moOpenAI claims maths breakthrough on a famed 'Millennium Problem'—
↗Not briefed
bluesky
bluesky-ai
2moI stopped reading at “ideas of their own”. An LLM has no more “ideas” than my sock.I stopped reading at “ideas of their own”. An LLM has no more “ideas” than my sock.
↗Not briefed
bluesky
bluesky-ai
2moAdobe’s gonna double down on Gen AI. Apparently they believe that people are dropping their products because creativity takes hard work & …Adobe’s gonna double down on Gen AI. Apparently they believe that people are dropping their products because creativity takes hard wo…
↗Not briefed
bluesky
bluesky-ai
2moi think "AI psychosis" actually has fuck all to do with AI it's just what has happened to dictators and CEOs for centuries, when you're su…i think "AI psychosis" actually has fuck all to do with AI it's just what has happened to dictators and CEOs for centuries, when you'…
↗Not briefed
bluesky
bluesky-ai
2moDid not wake up realizing that “did OpenAI cheat on the homework” would be the most exciting social media drama of the day cims.nyu.edu/~…Did not wake up realizing that “did OpenAI cheat on the homework” would be the most exciting social media drama of the day cims.nyu.…
↗Not briefed
bluesky
bluesky-ai
2moanyone who's impressed by "three OpenAI civilizations got started, then got wiped out, only to reemerge from the predecessor’s ashes" is go…anyone who's impressed by "three OpenAI civilizations got started, then got wiped out, only to reemerge from the predecessor’s ashes"…
↗Not briefed
hn
hn-ai
2moJava and Polyglot Microservice Benchmarks (K6 Load Test)—
↗Not briefed
bluesky
bluesky-ai
2mo“I asked whether the model had been trained on, or had access to, our sessions in Codex, into which we had been putting all our drafts for …“I asked whether the model had been trained on, or had access to, our sessions in Codex, into which we had been putting all our drafts…
↗Not briefed
arxiv
arxiv-ai
2moContractEval: Query-Conditioned Execution Matching for Procedural Instruction ConformanceAs LLM agents move from answering questions to carrying out procedures, failures can be unwarranted rather than visibly wrong: the fin…
↗Not briefed
bluesky
bluesky-ai
2moThe trick is not extrapolating from LLM accomplishments, because we are really quite awful at keeping the model of human reasoning we have …The trick is not extrapolating from LLM accomplishments, because we are really quite awful at keeping the model of human reasoning we…
↗Not briefed
bluesky
bluesky-ai
2moYeah, and there's a lot of unsettled debates in CogSci about why these disparities continue, but I think it's just a good reason to mentall…Yeah, and there's a lot of unsettled debates in CogSci about why these disparities continue, but I think it's just a good reason to me…
↗Not briefed
bluesky
bluesky-ai
2moAnthropic is no shining example of moral integrity, but it’s truly the lowest hanging fruit to look like the ethical/safe choice in this fi…Anthropic is no shining example of moral integrity, but it’s truly the lowest hanging fruit to look like the ethical/safe choice in th…
↗Not briefed
hn
hn-ai
2moAstra is fully rolled out to Plus, Pro, Business, and Enterprise users—
↗Not briefed
bluesky
bluesky-ai
2moCost them $15m in compute and lord knows how much electricity to do whatever shadiness they just did - and we still don’t know much about t…🔧 Hugging FaceCost them $15m in compute and lord knows how much electricity to do whatever shadiness they just did - and we still don’t know much ab…
↗Not briefed
bluesky
bluesky-ai
2moAuthors who are members of the class in the Anthropic Settlement, please be aware that if you have a work in the settlement which has multi…Authors who are members of the class in the Anthropic Settlement, please be aware that if you have a work in the settlement which has…
↗Not briefed
bluesky
bluesky-ai
2moMillennium math problem solved by OpenAI may have "regurgitated" competitors unpublished work... also they spent more than $22 million in c…Millennium math problem solved by OpenAI may have "regurgitated" competitors unpublished work... also they spent more than $22 million…
↗Not briefed
hn
hn-ai
2moShow HN: SimpleW – Lightweight C# Web Server with optional addonsHi HN,<p>I’m the author of SimpleW, an MIT-licensed .NET web server for standalone or embedded applications.<p>I shared it here last y…
↗Not briefed
arxiv
arxiv-ai
2moA Block Tensor Train Burer-Monteiro Framework for Low-Rank Quantum State TomographyQuantum state tomography is a fundamental technique for estimating the state of a quantum system from measured data and plays a crucia…
↗Not briefed
hn
hn-ai
2moMeta Just Launched MuseFew days ago Anthropic launched Claude Fable 5.1, then came OpenAI with Gpt-6 Astra and now Meta came up with Muse. The AGI competitio…
↗Not briefed
bluesky
bluesky-ai
2moHackers are stealing Claude tokens from subscribersHackers are stealing Claude tokens from subscribers
↗Not briefed
bluesky
bluesky-ai
2moon nie jest fanbojem, tylko od dawna jest finansowany przez openai czy kto tam odpowiada za sypanie hajsu na propagandęon nie jest fanbojem, tylko od dawna jest finansowany przez openai czy kto tam odpowiada za sypanie hajsu na propagandę
↗Not briefed