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
1moGeometric Mean Pooling for Equal-Weight Multiplicative Coarse-GrainingAs an alternative to the additive and extremal biases of average and max pooling, we introduce Geometric Mean Pooling (GMP), a signed…
↗Not briefed
bluesky
bluesky-ai
1moBuilt on theft is real, ongoing, and glided past by too many. This is an ILLEGAL enterprise. And the settlements, as with Anthropic and wri…Built on theft is real, ongoing, and glided past by too many. This is an ILLEGAL enterprise. And the settlements, as with Anthropic an…
↗Not briefed
bluesky
bluesky-ai
1monathan fielder should do a rehearsal season about the hugging face hack backofmind.substack.com/p/13-theses-...🔧 Hugging Facenathan fielder should do a rehearsal season about the hugging face hack backofmind.substack.com/p/13-theses-...
↗Not briefed
bluesky
bluesky-ai
1moFirefox using Mistral LLM is an interesting choice. I'm glad there is one switch to rule them all, but I'd feel better if that Block AI Enh…Firefox using Mistral LLM is an interesting choice. I'm glad there is one switch to rule them all, but I'd feel better if that Block A…
↗Not briefed
arxiv
arxiv-ai
1moChronosphere: Space-Time Tessellation of Local Climate ExpertsWe introduce Chronosphere, a spatio-temporal neural field that learns representations of climate. A central challenge in geographic re…
↗Not briefed
rss
wired-ai
1moAI PACs Have Dumped Nearly $1 Million Into an Obscure Senate RaceThe reliably Republican South Dakota senate seat has an incumbent on the ballot. But PACs associated with AI labs and investors have a…
↗Not briefed
rss
techcrunch-ai
1moRobinhood’s Abhishek Fatehpuria on winning the modern financial consumer at TechCrunch Disrupt 2026Robinhood’s Abhishek Fatehpuria on winning the modern financial consumer at TechCrunch Disrupt 2026. Register now to save up to $200 b…
↗Not briefed
rss
wired-ai
1moIf the AI Industry Followed Its Own Research, It Might Have Paused AlreadyAnthropic’s CEO says that safety hinges on understanding how AI “thinks.” So far the evidence is disturbing.
↗Not briefed
arxiv
arxiv-ai
1moNeural Cellular Automata Learn General Features in their Hidden Channels🔧 ConsensusModern deep learning models achieve impressive generalization through over-parameterization, but this paradigm often struggles with ov…
↗Not briefed
hn
hn-ai
1moShow HN: Omabib, a Zotero alternative for Omarchy with agent support—
↗Not briefed
bluesky
bluesky-ai
1moLess than a day until GPT-6 Astra is ready for more work! 😂 #AI #dev #Anthropic #Claude #ClaudeCode #OpenAI #ChatGPT #Codex #gaming #vide…🔧 ChatGPTLess than a day until GPT-6 Astra is ready for more work! 😂 #AI #dev #Anthropic #Claude #ClaudeCode #OpenAI #ChatGPT #Codex #gaming…
↗Not briefed
bluesky
bluesky-ai
1moY a un effet beaucoup plus pervers des LLM sur les metiers "à création argumentaire" (je prends ce terme pour integrer la basse bureaulerie…Y a un effet beaucoup plus pervers des LLM sur les metiers "à création argumentaire" (je prends ce terme pour integrer la basse bureau…
↗Not briefed
hn
hn-ai
1moShow HN: Rickub – The Smartest Git in the Universe—
↗Not briefed
bluesky
bluesky-ai
1moCoca-Cola's recipe is hidden in a vault, which seems to be a safer approach to whatever these AI companies are doing/building.... www.wsj.…Coca-Cola's recipe is hidden in a vault, which seems to be a safer approach to whatever these AI companies are doing/building.... www…
↗Not briefed
bluesky
bluesky-ai
1moA lot of people are reading the NYT’s filing in the NYT vs. OpenAI lawsuit like it’s some kind of gotcha and I would encourage you not to d…A lot of people are reading the NYT’s filing in the NYT vs. OpenAI lawsuit like it’s some kind of gotcha and I would encourage you not…
↗Not briefed
bluesky
bluesky-ai
1moA ver si hay suerte y se destruyen las unas a las otras, y de paso a sus dueños, antes de que lo hagan con la humanidad.A ver si hay suerte y se destruyen las unas a las otras, y de paso a sus dueños, antes de que lo hagan con la humanidad.
↗Not briefed
bluesky
bluesky-ai
1moOne of the reasons the murderer of their CEO claimed what he was doing was 'justice' is because of the many people who died because United…One of the reasons the murderer of their CEO claimed what he was doing was 'justice' is because of the many people who died because U…
↗Not briefed
rss
techmeme
1moMind, whose AI agents autonomously handle data loss prevention tasks, raised a $72M Series B, sources say at a $300M valuation, taking its total raised to $112M (Sophie Shulman/CTech)Sophie Shulman / CTech: Mind, whose AI agents autonomously handle data loss prevention tasks, raised a $72M Series B, sources say at a…
↗Not briefed
arxiv
arxiv-ai
1moAutoRecLab: Describe the Experiment, Get the Code!🔧 GPT-5Empirical evaluation is central to recommender-systems (RecSys) research, but turning experimental designs into executable code remain…
↗Not briefed
arxiv
arxiv-ai
1moTrialAtlas: Multi-Agent Research Organization for Clinical Trial Design and OptimizationNearly 90% of drugs entering clinical development ultimately fail, despite billions of dollars in investment. Pharmaceutical companies…
↗Not briefed
bluesky
bluesky-ai
1moJon Tester is on an Anthropic Council. Tester supports Bodnar. #MTPol www.anthropic.com/news/introdu...Jon Tester is on an Anthropic Council. Tester supports Bodnar. #MTPol www.anthropic.com/news/introdu...
↗Not briefed
arxiv
arxiv-ai
1moWatermarkable Multi-Draft Speculative Sampling via Poisson ProcessesLarge language models (LLMs) have achieved state-of-the-art performance across a wide range of tasks, motivating two important aspects…
↗Not briefed
arxiv
arxiv-ai
1moDo Personality-Tuned LLMs Make Better Social Agents?LLMs are increasingly used in social simulations for socially interactive agents and robots, offering more flexibility than rule-based…
↗Not briefed
bluesky
bluesky-ai
1moCommentary: Chinese open‑weight AI tools like Kimi and DeepSeek show that verifiable performance matters more than national origin.Commentary: Chinese open‑weight AI tools like Kimi and DeepSeek show that verifiable performance matters more than national origin.
↗Not briefed
bluesky
bluesky-ai
1moThis isn't a transparent effort to lean into the "AI can cure diseases!" argument that AI boosters always make when everyone is complaining…This isn't a transparent effort to lean into the "AI can cure diseases!" argument that AI boosters always make when everyone is compla…
↗Not briefed
hn
hn-ai
1moLet GPT-6 Astra code without using Codex Usage—
↗Not briefed
bluesky
bluesky-ai
1moAnthropic diz que IA avança para criar a si mesmaAnthropic diz que IA avança para criar a si mesma
↗Not briefed
bluesky
bluesky-ai
1mo➡️ L'intrusion, opérée par des chercheurs de la start-up de cybersécurité Hacktron AI, est intervenue fin juillet. Confirmée par OpenAI, el…➡️ L'intrusion, opérée par des chercheurs de la start-up de cybersécurité Hacktron AI, est intervenue fin juillet. Confirmée par OpenA…
↗Not briefed
hn
hn-ai
1moShow HN: Negronlyfans – upload your Negroni, get judged by its 1919 inventor—
↗Not briefed
hn
hn-ai
1moI built Anchor. The open-source ontology layer for AI agents—
↗Not briefed
bluesky
bluesky-ai
1moOpenAI's hiring of Jessica Schumer to their Northeast public policy team isn't just a major conflict of interest for the Senate Majority Le…OpenAI's hiring of Jessica Schumer to their Northeast public policy team isn't just a major conflict of interest for the Senate Majori…
↗Not briefed
arxiv
arxiv-ai
1moThe Weight Is Over - Interactive Diffusion on Consumer GPUsOn-device inference is booming, but the momentum is almost all in language models. Diffusion pipelines are memory hungry, latency-sens…
↗Not briefed
bluesky
bluesky-ai
1moThis seemed to fly under the radar when it came out last week, but it’s quite seismic: Anthropic is proactively monitoring dissent & report…This seemed to fly under the radar when it came out last week, but it’s quite seismic: Anthropic is proactively monitoring dissent & r…
↗Not briefed
arxiv
arxiv-ai
1moReusing Latent Speech Representations for Query-Conditioned Topic Localization in TranscriptsLong transcripts are costly inputs for downstream NLP systems and often contain irrelevant context. We study query-conditioned topic l…
↗Not briefed
bluesky
bluesky-ai
1moBasically, any LLM that is deployed at scale saws off the branch it's sitting on.Basically, any LLM that is deployed at scale saws off the branch it's sitting on.
↗Not briefed
bluesky
bluesky-ai
1moQuestion 1: yes, we do know. This is mere 'mirroring', in no sense comparable to a human learning things. See my recent posts on this topic…Question 1: yes, we do know. This is mere 'mirroring', in no sense comparable to a human learning things. See my recent posts on this…
↗Not briefed
arxiv
arxiv-ai
1moEnterpriseVal: Quantifying the Efficacy, Reliability and Value of Generative AI in the EnterpriseFrontier language models now produce professional deliverables that expert graders judge to match human work on a substantial share of…
↗Not briefed
bluesky
bluesky-ai
1moThis doom loop in relation to Canvas is also notable. OpenAI saw Canvas as a massive (and seemingly endless) trove of human text to mine f…This doom loop in relation to Canvas is also notable. OpenAI saw Canvas as a massive (and seemingly endless) trove of human text to m…
↗Not briefed
bluesky
bluesky-ai
1mo‘What kind of legal liability might an A.I. lab like OpenAI or Anthropic face if its products were to run amok and cause great harm?‘ www.n…‘What kind of legal liability might an A.I. lab like OpenAI or Anthropic face if its products were to run amok and cause great harm?‘…
↗Not briefed
bluesky
bluesky-ai
1moI've seen: automoderation, game playing, System 1 assessment of whether System 2 [your big LLM] should be invoked, general fun [applying it…I've seen: automoderation, game playing, System 1 assessment of whether System 2 [your big LLM] should be invoked, general fun [applyi…
↗Not briefed
bluesky
bluesky-ai
1moIt was good that nuclear technology was regulated. It is bad that AI is not. Still, I would not hold up the Atomic Energy Commission as a m…It was good that nuclear technology was regulated. It is bad that AI is not. Still, I would not hold up the Atomic Energy Commission a…
↗Not briefed
rss
simon-willison
1moThe Creative Spirit of Who Framed Roger Rabbit—
↗Not briefed
bluesky
bluesky-ai
1mo"Anthropic says... " stop right there, no thanks"Anthropic says... " stop right there, no thanks
↗Not briefed
bluesky
bluesky-ai
1mop-AI-ndora's box... What the %¨$*# have they created? #ai #artificialintelligence #nvidia #OpenAI #xAI #Meta #anthropic #Pandora @cartoonm…p-AI-ndora's box... What the %¨$*# have they created? #ai #artificialintelligence #nvidia #OpenAI #xAI #Meta #anthropic #Pandora @car…
↗Not briefed
bluesky
bluesky-ai
1mothis is an entirely new way of working, and companies should be preparing their employees to be able to answer questions like “is an LLM ev…this is an entirely new way of working, and companies should be preparing their employees to be able to answer questions like “is an L…
↗Not briefed
bluesky
bluesky-ai
1moNo wonder these AI companies are worrying about human extinction via AI created biological pathogens: They’re starting their own wet labs.No wonder these AI companies are worrying about human extinction via AI created biological pathogens: They’re starting their own wet…
↗Not briefed
rss
the-decoder
1moVisible chains of thought are a safety advantage for AI, but that transparency is slipping awayAI models think out loud today, but Google Deepmind says that transparency is at risk. The article Visible chains of thought are a saf…
↗Not briefed
bluesky
bluesky-ai
1moWhat if I told you AI was being used to kill people right now? arstechnica.com/health/2023/...What if I told you AI was being used to kill people right now? arstechnica.com/health/2023/...
↗Not briefed
bluesky
bluesky-ai
1moMaybe it's just because I'm a middle-aged millennial, but I do not yearn to chat with a sycophantic bot. I liked poking around the web. L…Maybe it's just because I'm a middle-aged millennial, but I do not yearn to chat with a sycophantic bot. I liked poking around the we…
↗Not briefed
bluesky
bluesky-ai
1moMientras no mande un mensaje "mata a Sarah Connor" a su versión pasada todo bajo control.Mientras no mande un mensaje "mata a Sarah Connor" a su versión pasada todo bajo control.
↗Not briefed