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
arxiv
arxiv-ai
2moOPD-Aha: From Linguistic Momentum to Visual Reflection in Multimodal On-Policy DistillationPrivileged on-policy distillation improves multimodal reasoning by allowing a teacher to evaluate student trajectories using rich, tra…
↗Not briefed
bluesky
bluesky-ai
2moI respect the Chinese AI companies more because they straight up say that the purpose of the technology is achieving communism. this is fro…I respect the Chinese AI companies more because they straight up say that the purpose of the technology is achieving communism. this i…
↗Not briefed
hn
hn-ai
2moAustralian insurer to recognise Tesla's FSD technology in calculating premiums—
↗Not briefed
arxiv
arxiv-ai
2moLanguage Orthogonalization for Zero-Shot Cross-Lingual Audio Deepfake DetectionAudio deepfake detectors need to transfer to languages absent from training, as multilingual speech synthesis outpaces labeled anti-sp…
↗Not briefed
hn
hn-ai
2moShow HN: Pi-box: Pi coding agent running inside a VM inside WASM inside ElectronProject background and human written summary here: <a href="https:&#x2F;&#x2F;deepclause.substack.com&#x2F;p&#x2F;pie-in-a-box-running…
↗Not briefed
bluesky
bluesky-ai
2modid anyone else think it was a little odd that the story of sam altman, amodei and elon musk agreeing about slowing down AI came out right …did anyone else think it was a little odd that the story of sam altman, amodei and elon musk agreeing about slowing down AI came out r…
↗Not briefed
hn
hn-ai
2moAI Agent Wallet Offboarding Checklist—
↗Not briefed
bluesky
bluesky-ai
2moOpenAI is reportedly training ChatGPT to sound less human-like and reduce sycophantic responses, according to 404 Media.🔧 ChatGPTOpenAI is reportedly training ChatGPT to sound less human-like and reduce sycophantic responses, according to 404 Media.
↗Not briefed
arxiv
arxiv-ai
2moFine-Tuning Fixes Mode Collapse and Over-Dispersion in LLMsRecent work by Doshi and Hauser (2024), Bisbee et al. (2024), and Xie et al. (2026) raises concerns that outputs from large language m…
↗Not briefed
hn
hn-ai
2moShow HN: Chess Cheater DetectorI&#x27;ve been playing with this idea to build an estimation of chess cheating likelihood from an exported chess game (PGN format).<p>…
↗Not briefed
bluesky
bluesky-ai
2modeep respect to the deepseek researchers but it seems like a glass house to throw stones from either waydeep respect to the deepseek researchers but it seems like a glass house to throw stones from either way
↗Not briefed
bluesky
bluesky-ai
2mo'The growing debate over the 'AIpocalypse' is also, and above all, about big money''The growing debate over the 'AIpocalypse' is also, and above all, about big money'
↗Not briefed
arxiv
arxiv-ai
2moEarly-Bird Decoding: Accelerating Diffusion LLMs with Learnable Block Sizes and Parallel SamplingDiffusion large language models (dLLMs) offer a promising parallel decoding paradigm as an alternative to autoregressive generation th…
↗Not briefed
arxiv
arxiv-ai
2moDecentralized Gossip Learning and Federated Averaging for Histopathology Image ClassificationBreast histopathology analysis increasingly relies on distributed learning because direct data pooling across institutions is often re…
↗Not briefed
arxiv
arxiv-ai
2moAdaptive Bayesian Partner Selection for Federated Clinical CentersFederated learning (FL) in healthcare faces pronounced heterogeneity and temporal concept drift across clinical centers, where evolvin…
↗Not briefed
hn
hn-ai
2moOpenAI's malicious bot swarm attacked RubyGems—
↗Not briefed
hn
hn-ai
2moMouse on frontier harness with glm-5.3-flash—
↗Not briefed
bluesky
bluesky-ai
2moMETR vibe-coded a website and expose an API key that gave an attacker access to $600k worth of compute. But don't worry, METR has been nam…METR vibe-coded a website and expose an API key that gave an attacker access to $600k worth of compute. But don't worry, METR has bee…
↗Not briefed
hn
hn-ai
2moClaude for Financial Advisors—
↗Not briefed
bluesky
bluesky-ai
2moAnthropic researchers say AI could cause human extinction by 2030. Trump says Ehhhh, I'll lilkey be dead by then anyway. #Whatever www.theg…Anthropic researchers say AI could cause human extinction by 2030. Trump says Ehhhh, I'll lilkey be dead by then anyway. #Whatever www…
↗Not briefed
bluesky
bluesky-ai
2moMoreover, Chinese LLM companies and researchers actually document and share key insights of LLM design, as well as model weights, so others…Moreover, Chinese LLM companies and researchers actually document and share key insights of LLM design, as well as model weights, so o…
↗Not briefed
bluesky
bluesky-ai
2moThe physics that inspired neural networks didn’t stop being physics when it helped model memory and build thinking machines. https://www.qu…The physics that inspired neural networks didn’t stop being physics when it helped model memory and build thinking machines. https://w…
↗Not briefed
bluesky
bluesky-ai
2moThe AI industry perfectly encapsulated. "We're going to make a profit!!!!*" . .. . . . . . * if you exclude all our costs of doing business…The AI industry perfectly encapsulated. "We're going to make a profit!!!!*" . .. . . . . . * if you exclude all our costs of doing bus…
↗Not briefed
bluesky
bluesky-ai
2moでも、トランプは現状ではAI規制をやるつもりはない。 2024大統領選では、バイデンAI規制案に反発したアンドリーセンが主導し、多くのテック企業トップが一気にトランプ/共和党支持に転向した。 トランプが彼らの経済的・言論的支援(特にSNSでの世論形成)に恩義を感じているかど…でも、トランプは現状ではAI規制をやるつもりはない。 2024大統領選では、バイデンAI規制案に反発したアンドリーセンが主導し、多くのテック企業トップが一気にトランプ/共和党支持に転向した。 トランプが彼らの経済的・言論的支援(特にSNSでの世論形成)に恩義を感じ…
↗Not briefed
bluesky
bluesky-ai
2moWhat if your credit card rewarded you with AI tokens instead of airline miles? In China, that’s already happening, alongside token plans fr…What if your credit card rewarded you with AI tokens instead of airline miles? In China, that’s already happening, alongside token pla…
↗Not briefed
bluesky
bluesky-ai
2moMany posts about #ai slowdowns, on slashdot, hit the nail on the head. LLM companies overpromised and still haven't turned a profit. Feari…Many posts about #ai slowdowns, on slashdot, hit the nail on the head. LLM companies overpromised and still haven't turned a profit.…
↗Not briefed
bluesky
bluesky-ai
2moMusk Drops Apple Lawsuit Over ChatGPT Siri Deal, Keeps Suing OpenAI🔧 ChatGPTMusk Drops Apple Lawsuit Over ChatGPT Siri Deal, Keeps Suing OpenAI
↗Not briefed
hn
hn-ai
2moOpen-source Python framework for AI agents, workflows, and automations—
↗Not briefed
bluesky
bluesky-ai
2moおそらくどっちもあると思う。 2024年のバイデンのAI規制案に反旗を翻していた他のテック企業トップと違い、サム・アルトマンは微温的な態度を取ってたんだけど、それはOpenAIが確実に「政府のAI規制で護られる数社の側」に入れると見込んでいたから。 核開発におけるNPT(核…おそらくどっちもあると思う。 2024年のバイデンのAI規制案に反旗を翻していた他のテック企業トップと違い、サム・アルトマンは微温的な態度を取ってたんだけど、それはOpenAIが確実に「政府のAI規制で護られる数社の側」に入れると見込んでいたから。 核開発における…
↗Not briefed
arxiv
arxiv-ai
2moThe Neverwhere Visual Parkour Benchmark SuiteState-of-the-art visual locomotion controllers are increasingly capable at handling complex visual environments, making evaluating the…
↗Not briefed
rss
techmeme
2moApple's H1 2027 plans reportedly include an iPhone 18e, A20 iPhone 18, A20 Pro iPhone Air 2, OLED iPad Air, A19 Pro MacBook Neo, and M6 MacBook Air (Mark Gurman/Bloomberg)Mark Gurman / Bloomberg: Apple's H1 2027 plans reportedly include an iPhone 18e, A20 iPhone 18, A20 Pro iPhone Air 2, OLED iPad Air, A…
↗Not briefed
bluesky
bluesky-ai
2moもうひとつは、 ②Anthropicを含む米国系のAIベンチャーやビッグテック数社がフロンティアモデルを寡占する現状で、新たなAI開発に対して政府規制がかかれば、自分たちは(蒸留を交えて)キャッチアップしてくる後発組とのコスト競争を強いられずに、会員制の「AIクラブ」のラウン…もうひとつは、 ②Anthropicを含む米国系のAIベンチャーやビッグテック数社がフロンティアモデルを寡占する現状で、新たなAI開発に対して政府規制がかかれば、自分たちは(蒸留を交えて)キャッチアップしてくる後発組とのコスト競争を強いられずに、会員制の「AIクラブ…
↗Not briefed
bluesky
bluesky-ai
2moThe machines are here. John Connor knew. Ex-Anthropic Employee: Here’s Why A.I. Might Want to Kill Us www.nytimes.com/video/podcas...The machines are here. John Connor knew. Ex-Anthropic Employee: Here’s Why A.I. Might Want to Kill Us www.nytimes.com/video/podcas...
↗Not briefed
bluesky
bluesky-ai
2mois it acceptable to call into work sick when you run out of LLM usage tokens? y/nis it acceptable to call into work sick when you run out of LLM usage tokens? y/n
↗Not briefed
bluesky
bluesky-ai
2mo"I don't think we have a lot of time...Whatever we do, it has to happen fast...I would really target 2025, 2026, maybe even some chance of …"I don't think we have a lot of time...Whatever we do, it has to happen fast...I would really target 2025, 2026, maybe even some chanc…
↗Not briefed
bluesky
bluesky-ai
2moSo the Anthropic CEO is now using the "AI will become conscious and kill us" panic to start laying the foundations of justifying a Governme…So the Anthropic CEO is now using the "AI will become conscious and kill us" panic to start laying the foundations of justifying a Gov…
↗Not briefed
bluesky
bluesky-ai
2mohuggingface: openai:huggingface: openai:
↗Not briefed
bluesky
bluesky-ai
2moアモデイのこういう発言については複数の見方ができると思う。 ①アモデイは高校時代からEA(効果的利他主義)的な関心を持ち、ベンチャーやる前に理系研究者としてそこそこのキャリアを積んできた、AIテックのトップのなかでは倫理的側面に対する関心が高い人物ではある。10年前の以下論文…アモデイのこういう発言については複数の見方ができると思う。 ①アモデイは高校時代からEA(効果的利他主義)的な関心を持ち、ベンチャーやる前に理系研究者としてそこそこのキャリアを積んできた、AIテックのトップのなかでは倫理的側面に対する関心が高い人物ではある。10年前…
↗Not briefed
hn
hn-ai
2moHands-On Learning in the Age of AI—
↗Not briefed
bluesky
bluesky-ai
2mocredit to open ai for finally building a model that won't just agree with you when you tell it it's blatantly wrongcredit to open ai for finally building a model that won't just agree with you when you tell it it's blatantly wrong
↗Not briefed
hn
hn-ai
2moAsk HN: The amount of AI bot traffic is out control?We host a web app which has around 50,000 statically generated public pages, and the amount of bot traffic is insane. Facebook’s crawl…
↗Not briefed
bluesky
bluesky-ai
2moあぁ…米国のトランプ大統領が発言した内容は殆ど意味無いから無視して良いんじゃないですかね 司法省動かしてまでLLM開発止めさせたく無い理由なんて金以外にないでしょう 国の行く末よりも自分の懐に入る金銭に執着するような奴の評価を言葉通り受け取っちゃいけません。あぁ…米国のトランプ大統領が発言した内容は殆ど意味無いから無視して良いんじゃないですかね 司法省動かしてまでLLM開発止めさせたく無い理由なんて金以外にないでしょう 国の行く末よりも自分の懐に入る金銭に執着するような奴の評価を言葉通り受け取っちゃいけません。
↗Not briefed
arxiv
arxiv-ai
2moLearned Look-Ahead Splitting Rule for CARTClassification and regression trees are typically constructed using a greedy splitting rule that maximizes the immediate reduction in…
↗Not briefed
bluesky
bluesky-ai
2mo‘Just Common Sense’: Anthropic Co-Founder Goes on Fox News, Disagrees With Trump For Dismissing AI Fears‘Just Common Sense’: Anthropic Co-Founder Goes on Fox News, Disagrees With Trump For Dismissing AI Fears
↗Not briefed
bluesky
bluesky-ai
2moMe several AI generations later..."Thrice Claude Ceasar refused the crown!"Me several AI generations later..."Thrice Claude Ceasar refused the crown!"
↗Not briefed
bluesky
bluesky-ai
2moLLMはサボるし劣化するし、人間の監視なしに作業を任せてはならない|地雷魚 note.com/jiraygyo/n/n... “ LLM(大規模言語モデル)は使えば便利なツールだが、それはあくまで人間のチェックがあって成立するもので人間を代替する「夢の技術」などではない。 …LLMはサボるし劣化するし、人間の監視なしに作業を任せてはならない|地雷魚 note.com/jiraygyo/n/n... “ LLM(大規模言語モデル)は使えば便利なツールだが、それはあくまで人間のチェックがあって成立するもので人間を代替する「夢の技術」などで…
↗Not briefed
hn
hn-ai
2moAnthropic is in regulatory-capture financial loop—
↗Not briefed
arxiv
arxiv-ai
2moInterpreting and Steering LLM Agents for Social SimulationsSimulations based on large language models (LLMs) have proven to be powerful for understanding human behavior, making them valuable ad…
↗Not briefed
rss
bing-news-ai
2moSwiss SMEs Do Not View Artificial Intelligence as a Job KillerArtificial intelligence is now widely used among Swiss SMEs, but so far it has brought about few fundamental changes. Three-quarters o…
↗Not briefed
rss
bing-news-ai
2moSwiss SMEs Do Not View Artificial Intelligence as a Job KillerArtificial intelligence is now widely used among Swiss SMEs, but so far it has brought about few fundamental changes. Three-quarters o…
↗Not briefed