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
2moInoculation Midtraining with Learned NeologismsLarge language models (LLMs) often learn both desirable and undesirable properties during post-training. We study whether midtraining,…
↗Not briefed
bluesky
bluesky-ai
2moAnthropic is closing a multibillion-dollar deal with RUM, the parent company of the Trump-aligned online platform Rumble. All the tech bro…Anthropic is closing a multibillion-dollar deal with RUM, the parent company of the Trump-aligned online platform Rumble. All the tec…
↗Not briefed
arxiv
arxiv-ai
2moPrivacy-enhanced federated learning via asynchronous aggregation and local differential perturbationThis study proposes a privacy-enhanced federated learning framework to address secure collaborative training in distributed data envir…
↗Not briefed
bluesky
bluesky-ai
2moEven if you really feel like you need an LLM to rewrite your email I don't think anyone needs a 500 billion parameter model to do it. A mod…Even if you really feel like you need an LLM to rewrite your email I don't think anyone needs a 500 billion parameter model to do it.…
↗Not briefed
bluesky
bluesky-ai
2moRight. The strong and smart (High IQ!) President has AI all under control. The press should ask him what "LLM" stands for, or what "alignm…Right. The strong and smart (High IQ!) President has AI all under control. The press should ask him what "LLM" stands for, or what "a…
↗Not briefed
bluesky
bluesky-ai
2moAI boosters take it for granted that the best business model is "flood the market with spam." But it isn't! That's actually a very marginal…AI boosters take it for granted that the best business model is "flood the market with spam." But it isn't! That's actually a very mar…
↗Not briefed
arxiv
arxiv-ai
2moThin-shell stability of Gaussian cooling: logconcave sampling with sesteric complexity from a cold startWe show that logconcave probability measures along the Gaussian cooling path have thin-shell stability, generalizing the thin-shell th…
↗Not briefed
rss
techcrunch-ai
2moWith iOS 27, I’m actually using Siri againApple’s long-delayed Siri overhaul is finally here with iOS 27, and it changes how useful the assistant feels day to day
↗Not briefed
arxiv
arxiv-ai
2moLearning Multimodal One-step Flow Policy via Value-weighted Optimal TransportOffline reinforcement learning aims to learn a policy solely from fixed datasets, which often contain multimodal action distributions.…
↗Not briefed
bluesky
bluesky-ai
2moTech stocks slide on AI slowdown talks reut.rs/46WOn4HTech stocks slide on AI slowdown talks reut.rs/46WOn4H
↗Not briefed
hn
hn-ai
2moSixSentences – self-hostable workspace for the whole research process—
↗Not briefed
bluesky
bluesky-ai
2moOh hell nahhhh I just found a VTuber who very clearly uses AI while I was looking something up. Saw they've a few times used GENERATED imag…Oh hell nahhhh I just found a VTuber who very clearly uses AI while I was looking something up. Saw they've a few times used GENERATED…
↗Not briefed
bluesky
bluesky-ai
2moI wonder if people realise that uploading someone’s work into an “AI detector” just to try and accuse them of using an LLM to write it ulti…I wonder if people realise that uploading someone’s work into an “AI detector” just to try and accuse them of using an LLM to write it…
↗Not briefed
bluesky
bluesky-ai
2moIf this knucklehead says AI is safe, then you know it's not! abcnews.com/Politics/ope...If this knucklehead says AI is safe, then you know it's not! abcnews.com/Politics/ope...
↗Not briefed
bluesky
bluesky-ai
2moi found out one of my profs is putting LLM traps like "mention umbrellas" in tiny white font at the end of questions (because i copy paste …i found out one of my profs is putting LLM traps like "mention umbrellas" in tiny white font at the end of questions (because i copy p…
↗Not briefed
bluesky
bluesky-ai
2mowww.theguardian.com/business/202... This maniac is putting mankind in mortal danger. Will no one stop him ?www.theguardian.com/business/202... This maniac is putting mankind in mortal danger. Will no one stop him ?
↗Not briefed
bluesky
bluesky-ai
2moAnthropic confirmed it is doing much the same thing too, and is using human reviewers www.404media.co/inside-proje...Anthropic confirmed it is doing much the same thing too, and is using human reviewers www.404media.co/inside-proje...
↗Not briefed
hn
hn-ai
2moNCP-ArchPreview: 8.9B latent LM matches OLMo-3-7B on 51% of tokens—
↗Not briefed
hn
hn-ai
2moPatients Who Fight Health Insurance Denials Often Win—
↗Not briefed
bluesky
bluesky-ai
2moHashtagalignmentHashtagalignment
↗Not briefed
bluesky
bluesky-ai
2moIt is interesting how this thread immediately pivots into "actually AI doesn't work it's all mechanical turk" rather than the underlying pr…It is interesting how this thread immediately pivots into "actually AI doesn't work it's all mechanical turk" rather than the underlyi…
↗Not briefed
bluesky
bluesky-ai
2moChina's DeepSeek hires dealmaker as CFO ahead of possible IPO reut.rs/3T3g48EChina's DeepSeek hires dealmaker as CFO ahead of possible IPO reut.rs/3T3g48E
↗Not briefed
rss
techcrunch-ai
2moFashion app Daydream uses Apple Intelligence to help you shop the outfits in your camera rollThanks to the launch of iOS 27, Daydream's app now includes features that can turn saved outfit photos into shoppable results and sear…
↗Not briefed
arxiv
arxiv-ai
2moLLM-Based Schema-Aware Split Learning for Privacy-Preserving Mental Distress Prediction Across Heterogeneous SurveysRising societal and lifestyle complexity has been linked to a growing prevalence of mental distress worldwide. Educational institution…
↗Not briefed
hn
hn-ai
2moEU to limit access to social media before age of 15—
↗Not briefed
bluesky
bluesky-ai
2moI will say that, while I expect this to change over a long period of time, I could personally still get by fine without an LLM at work and …🔧 ChatGPTI will say that, while I expect this to change over a long period of time, I could personally still get by fine without an LLM at work…
↗Not briefed
bluesky
bluesky-ai
2moBREAKING: NVIDIA, Palantir, Booz Allen Hamilton and the Pentagon have said will limit their use of Anthropic models.BREAKING: NVIDIA, Palantir, Booz Allen Hamilton and the Pentagon have said will limit their use of Anthropic models.
↗Not briefed
hn
hn-ai
2moLimen: One-human-many-agents harness built from files, Git and one CLI—
↗Not briefed
bluesky
bluesky-ai
2moLa industria de la IA ‘mintió’ sobre los riesgos de la IA, dice el gran líder de la industria de la IA. Hay que tener la cara de cemento. …La industria de la IA ‘mintió’ sobre los riesgos de la IA, dice el gran líder de la industria de la IA. Hay que tener la cara de ceme…
↗Not briefed
bluesky
bluesky-ai
2mothe "heh, once all the LLM tech debt catches up and everyone realizes just how bad AI is I'll be there to clean it up with my manual handwr…the "heh, once all the LLM tech debt catches up and everyone realizes just how bad AI is I'll be there to clean it up with my manual h…
↗Not briefed
bluesky
bluesky-ai
2moCommentary: Anyone Else Have Those Weird Dreams Where Sobbing Future Generations Beg You To Change Course? By Sam Altman, CEO, OpenAI theon…Commentary: Anyone Else Have Those Weird Dreams Where Sobbing Future Generations Beg You To Change Course? By Sam Altman, CEO, OpenAI…
↗Not briefed
bluesky
bluesky-ai
2moAmazingAmazing
↗Not briefed
hn
hn-ai
2moPush-based vs. Pull-based Customization—
↗Not briefed
hn
hn-ai
2moA1ex: A simple LLM coding agent in Lua—
↗Not briefed
arxiv
arxiv-ai
2moLongAgent: History-Guided Agentic Search for Longitudinal Outcome PredictionExtracting informative representations from longitudinal data that can predict future outcomes remains a critical challenge in medicin…
↗Not briefed
hn
hn-ai
2moPrompt steering can reduce DeepSeek v4 hallucinations to below GPT and Claude—
↗Not briefed
bluesky
bluesky-ai
2moUS judge says she will stay on OpenAI case after divesting Microsoft stock reut.rs/4j3C3qEUS judge says she will stay on OpenAI case after divesting Microsoft stock reut.rs/4j3C3qE
↗Not briefed
arxiv
arxiv-ai
2moTask-Directed Residual AddUNet:Perfect-Reconstruction Routing for Full-Rate RepresentationsThis paper establishes a perfect-reconstruction (PR) interpretation of AddUNet and its full-rate realization, and introduces a Residua…
↗Not briefed
rss
wired-ai
2moNew York Seizes a Dozen Celebrity Deepfake WebsitesIn the biggest-ever legal action against harmful deepfake websites, the Manhattan District Attorney’s Office has seized 12 sites that…
↗Not briefed
arxiv
arxiv-ai
2moK-Bench: a clinically calibrated benchmark for evaluating large language models in high-risk mental health conversations🔧 Consensus% !TEX root = ../main.tex People increasingly use large language models (LLMs) for mental health support, yet their safety in evolving…
↗Not briefed
bluesky
bluesky-ai
2mohuh who wouldve guessed a llm company would allow third parties to view sensitive private information without clear consent www.404media.c…huh who wouldve guessed a llm company would allow third parties to view sensitive private information without clear consent www.404me…
↗Not briefed
bluesky
bluesky-ai
2moBlack tshirt boy 🖤 #boy #man #hot #gay #sex #model #ai #ia #gymboy #hotboy #nudeboy #nude #guy #hunk #twink #twunk #sexyboy #dude #gooner …Black tshirt boy 🖤 #boy #man #hot #gay #sex #model #ai #ia #gymboy #hotboy #nudeboy #nude #guy #hunk #twink #twunk #sexyboy #dude #go…
↗Not briefed
bluesky
bluesky-ai
2moThe dumbest person on the planet says that his intelligence alone is enough to keep AI from becoming dangerous. Dunning-Kruger doesn't eve…The dumbest person on the planet says that his intelligence alone is enough to keep AI from becoming dangerous. Dunning-Kruger doesn'…
↗Not briefed
bluesky
bluesky-ai
2mowe're all really great at roasting Raymond and not a one of us has needed an LLM to do itwe're all really great at roasting Raymond and not a one of us has needed an LLM to do it
↗Not briefed
bluesky
bluesky-ai
2moI'm a team lead, so I actually do care how it was done: if they used an LLM, I can't trust that they learned anything from doing the work …I'm a team lead, so I actually do care how it was done: if they used an LLM, I can't trust that they learned anything from doing the…
↗Not briefed
bluesky
bluesky-ai
2moRedhair boy 🧡 #boy #man #hot #gay #sex #model #ai #ia #gymboy #hotboy #nudeboy #nude #guy #hunk #twink #twunk #sexyboy #dude #gooner #gaym…Redhair boy 🧡 #boy #man #hot #gay #sex #model #ai #ia #gymboy #hotboy #nudeboy #nude #guy #hunk #twink #twunk #sexyboy #dude #gooner…
↗Not briefed
bluesky
bluesky-ai
2moMonster cock boy 🍆 #boy #man #hot #gay #sex #model #ai #ia #gymboy #hotboy #nudeboy #nude #guy #hunk #twink #twunk #sexyboy #dude #gooner …Monster cock boy 🍆 #boy #man #hot #gay #sex #model #ai #ia #gymboy #hotboy #nudeboy #nude #guy #hunk #twink #twunk #sexyboy #dude #go…
↗Not briefed
arxiv
arxiv-ai
2moLearning to Coach for Experiential LearningLanguage models can learn from experience, but raw solution trajectories are often too long and noisy to provide effective guidance. I…
↗Not briefed
bluesky
bluesky-ai
2moThe "security profiles" that are literally just anthropic asking Claude to review chat logs, lolThe "security profiles" that are literally just anthropic asking Claude to review chat logs, lol
↗Not briefed
bluesky
bluesky-ai
2moLove that during the High Holidays I got to explain LessWrong lore, Aella, and what X risk are to octogenarian Jewish seniors. Since of co…Love that during the High Holidays I got to explain LessWrong lore, Aella, and what X risk are to octogenarian Jewish seniors. Since…
↗Not briefed