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
1moFrom over on X, some more context around the CNC parties in the Silicon Valley/Rationalist/AGI set:From over on X, some more context around the CNC parties in the Silicon Valley/Rationalist/AGI set:
↗Not briefed
arxiv
arxiv-ai
1moDiaVLo: Diagnosing Behaviours of Vision-Language ModelsVision-language models (VLMs) rely on storing and transferring appropriate information across their sub-components. Verifying that the…
↗Not briefed
rss
techcrunch-ai
1moAutomattic’s 33-Hour Coup, and can AI labs police themselves?A week after an Anthropic researcher’s doomsday warning rattled the AI world, the company’s CEO Dario Amodei has outlined his plan to…
↗Not briefed
bluesky
bluesky-ai
1moAir gapping apparently wouldn't work in alignment tests because, I shit you not according to this OpenAI guy, one computer could run the CP…Air gapping apparently wouldn't work in alignment tests because, I shit you not according to this OpenAI guy, one computer could run t…
↗Not briefed
bluesky
bluesky-ai
1moopenAI and microsoft just got four strikes from bluesky moderationopenAI and microsoft just got four strikes from bluesky moderation
↗Not briefed
bluesky
bluesky-ai
1molike the most interesting models with the most actual-use potential right now, for my money, are GLM-5.3 and GLM-5.3 Flash, with Deepseek 4…like the most interesting models with the most actual-use potential right now, for my money, are GLM-5.3 and GLM-5.3 Flash, with Deeps…
↗Not briefed
rss
theverge-ai
1moGavin Newsom is pushing for an AI kill switchCalifornia Gov. Gavin Newsom (D) is positioning the state to take the lead on AI oversight, including the potential to mandate a "kill…
↗Not briefed
bluesky
bluesky-ai
1mo2) The "we accidentally made the Terminator" messaging is about forcing government regulation. Why would they want regulation? Because it…2) The "we accidentally made the Terminator" messaging is about forcing government regulation. Why would they want regulation? Becau…
↗Not briefed
hn
hn-ai
1moGLM-5.3-FlashX: Delivering inference speeds of 200 tokens/s—
↗Not briefed
arxiv
arxiv-ai
1moAbstention and Noise Filtering: Two Missing Primitives of Softmax AttentionGating the value pathway of attention reportedly improves language model pretraining, and prior studies disagree on why. We argue and…
↗Not briefed
hn
hn-ai
1moChinese AI Radicalizes—
↗Not briefed
bluesky
bluesky-ai
1moJev looks really cool. a way to leverage techniques from generative AI that isn't actually *generating* anything, just classifying or answe…Jev looks really cool. a way to leverage techniques from generative AI that isn't actually *generating* anything, just classifying or…
↗Not briefed
bluesky
bluesky-ai
1moThis may be a little awkward, but this is why I think the discussion around "AI doomerism" is kinda broken.. Imagine a political compass. …This may be a little awkward, but this is why I think the discussion around "AI doomerism" is kinda broken.. Imagine a political comp…
↗Not briefed
arxiv
arxiv-ai
1moRecreationWorld: Scalable and Verifiable Environments for Hybrid Computer-Use AgentsComputer-use agents (CUAs) have advanced along two separate lines: graphical interaction and software development through code and the…
↗Not briefed
rss
stratechery
1mo2026.38: DoomforceThe best Stratechery content from the week of September 14, 2026, including the view from anywhere but San Francisco, the limited pote…
↗Not briefed
rss
bing-news-ai
1moBetter Artificial Intelligence Stock: Aehr Test Systems vs. Super Micro ComputerAehr trades at a steep premium despite recent losses, while Super Micro's valuation looks cheap relative to its growth, but both carry…
↗Not briefed
rss
bing-news-ai
1moBetter Artificial Intelligence Stock: Aehr Test Systems vs. Super Micro ComputerAehr trades at a steep premium despite recent losses, while Super Micro's valuation looks cheap relative to its growth, but both carry…
↗Not briefed
rss
bing-news-ai
1moBetter Artificial Intelligence Stock: Aehr Test Systems vs. Super Micro ComputerAehr trades at a steep premium despite recent losses, while Super Micro's valuation looks cheap relative to its growth, but both carry…
↗Not briefed
rss
bing-news-ai
1moArtificial Intelligence and Green City Transition Forum successfully held at GreenCities 2026On September 15th, Artificial Intelligence and Green City Transition Innovation Forum was successfully held in GreenCities 2026. The f…
↗Not briefed
rss
bing-news-ai
1moBetter Artificial Intelligence Stock: Aehr Test Systems vs. Super Micro ComputerAehr trades at a steep premium despite recent losses, while Super Micro's valuation looks cheap relative to its growth, but both carry…
↗Not briefed
rss
bing-news-ai
1moArtificial Intelligence and Green City Transition Forum successfully held at GreenCities 2026On September 15th, Artificial Intelligence and Green City Transition Innovation Forum was successfully held in GreenCities 2026. The f…
↗Not briefed
bluesky
bluesky-ai
1moOpenAI breached via a Discourse RCE Hackers accessed OpenAI's internal codebase and employee accounts through an RCE in their forum softwa…OpenAI breached via a Discourse RCE Hackers accessed OpenAI's internal codebase and employee accounts through an RCE in their forum s…
↗Not briefed
bluesky
bluesky-ai
1moI wanted to break away from the hype/doom cycle surrounding AI companies and their strange new interest in famous open problems, so I tried…I wanted to break away from the hype/doom cycle surrounding AI companies and their strange new interest in famous open problems, so I…
↗Not briefed
hn
hn-ai
1moTickrs: Realtime stock ticker data in your terminal—
↗Not briefed
bluesky
bluesky-ai
1moI see you wrote “hypnovisor” but what I read was “hypervisor” and I have bad news, they’re running docker containers on your brainmeats now…I see you wrote “hypnovisor” but what I read was “hypervisor” and I have bad news, they’re running docker containers on your brainmeat…
↗Not briefed
arxiv
arxiv-ai
1moBayesian Belief Layer for Controllable Opinion Dynamics in LLM Agents🔧 ConsensusLLM agents in social simulation revise their opinions implicitly, in context: how open an agent is to persuasion can neither be specif…
↗Not briefed
bluesky
bluesky-ai
1moOh no I agree with all that. I thought you meant the apocalyptic doomerism where they imagine LLM chatbots physically killing all of us wit…Oh no I agree with all that. I thought you meant the apocalyptic doomerism where they imagine LLM chatbots physically killing all of u…
↗Not briefed
arxiv
arxiv-ai
1moA Lie Detector Test for Language Models: Reading Knowledge a Model Won't RevealLarge language models can hold knowledge they do not report. A model may sandbag on a capability evaluation, or answer against what it…
↗Not briefed
arxiv
arxiv-ai
1moAssessment of Machine Learning-Based Critical Heat Flux Models in the CTF Subchannel Code for Square Rod Bundle Prediction🔧 FLUXThe prediction of critical heat flux (CHF), a key safety-related quantity in nuclear thermal hydraulics, remains an important challeng…
↗Not briefed
bluesky
bluesky-ai
1mobut because so many reporters have had their brains utterly cooked by twitter and elons algorithm amplifying a bunch of bot farm accounts s…but because so many reporters have had their brains utterly cooked by twitter and elons algorithm amplifying a bunch of bot farm accou…
↗Not briefed
hn
hn-ai
1moShow HN: A Prometheus exporter for Sagemcom F3896 cable modems—
↗Not briefed
bluesky
bluesky-ai
1moOh dear, calamitously inaccurate #ornithology science communication brought to you by an AI text-to-image model. via @sharpebirder.bsky.so…Oh dear, calamitously inaccurate #ornithology science communication brought to you by an AI text-to-image model. via @sharpebirder.bs…
↗Not briefed
arxiv
arxiv-ai
1moMoral Entropy: Auditing Bias and Uncertainty in Moral Judgment🔧 ConsensusMost work in computational ethics treats annotator disagreement on moral content as noise to be voted away, collapsed into majority vo…
↗Not briefed
arxiv
arxiv-ai
1moTime series generation with spectrally aligned latent flow matchingLatent flow models have proven to be a reliable and cost-effective method for time series generation. However, the latent compression…
↗Not briefed
bluesky
bluesky-ai
1moI will never knowingly stream a game that uses GenAI/LLM content. I've been offered review copies of such games on two occasions and reject…I will never knowingly stream a game that uses GenAI/LLM content. I've been offered review copies of such games on two occasions and r…
↗Not briefed
hn
hn-ai
1moShare an LLM endpoint over HTTPS while TLS still terminates on your machine—
↗Not briefed
bluesky
bluesky-ai
1moI'm so tired of AI doomer rhetoric because it makes worrying about AI sound like a crazy person thing. No, I am not worried about Claude p…I'm so tired of AI doomer rhetoric because it makes worrying about AI sound like a crazy person thing. No, I am not worried about Cla…
↗Not briefed
bluesky
bluesky-ai
1moYou know, whenever I see articles going off about hacking, especially within LLM shops, I look at my cybersecurity designations and wonder …You know, whenever I see articles going off about hacking, especially within LLM shops, I look at my cybersecurity designations and wo…
↗Not briefed
bluesky
bluesky-ai
1mo“An astonishing theft of unprecedented proportions” and the “largest theft of labor in human history” is what a Microsoft executive private…“An astonishing theft of unprecedented proportions” and the “largest theft of labor in human history” is what a Microsoft executive pr…
↗Not briefed
bluesky
bluesky-ai
1mohere's the core idea: LLMs have all that crazy haskell type level DSL magic in their weights, know it from token 0. Therefore, Haskell is a…here's the core idea: LLMs have all that crazy haskell type level DSL magic in their weights, know it from token 0. Therefore, Haskell…
↗Not briefed
rss
techmeme
1moGoogle threat intelligence group details how one of its researchers infiltrated hacker group TeamPCP and helped disrupt its software supply chain hacking spree (Andy Greenberg/Wired)Andy Greenberg / Wired: Google threat intelligence group details how one of its researchers infiltrated hacker group TeamPCP and helpe…
↗Not briefed
hn
hn-ai
1moShow HN: Jev helps you to not run malicous code—
↗Not briefed
bluesky
bluesky-ai
1moThere's just no world where Anthropic and OpenAI aren't looking at Jev and being like, "oh yeah we can make that too" and then just making …There's just no world where Anthropic and OpenAI aren't looking at Jev and being like, "oh yeah we can make that too" and then just ma…
↗Not briefed
hn
hn-ai
1moMy Thoughts on AI and LLMsAI is already superhuman in some aspects, it knows way more information than a single human can hold. What it lacks is what makes us h…
↗Not briefed
bluesky
bluesky-ai
1moA Microsoft executive has called AI training practices "the largest theft of labor in human history," according to new files in our lawsuit…A Microsoft executive has called AI training practices "the largest theft of labor in human history," according to new files in our la…
↗Not briefed
bluesky
bluesky-ai
1moOpenAI And Microsoft Knew Their Sh*t Would Kill Newspapers Like Big Tobacco Knew About Lung Cancer And Big Oil Knew About Climate Change an…OpenAI And Microsoft Knew Their Sh*t Would Kill Newspapers Like Big Tobacco Knew About Lung Cancer And Big Oil Knew About Climate Chan…
↗Not briefed
bluesky
bluesky-ai
1moThis court filing seeking summary judgment by The NYT and other news organizations in the copyright infringement lawsuit against OpenAI is …This court filing seeking summary judgment by The NYT and other news organizations in the copyright infringement lawsuit against OpenA…
↗Not briefed
arxiv
arxiv-ai
1moMultiplicative Optimism for Constant Regret in GamesWe introduce Multiplicatively Optimistic Regret Matching (MORM), an uncoupled learning rule for finite general-sum games. Under simult…
↗Not briefed
bluesky
bluesky-ai
1moespecially when we have prominent rats and EAs being absolute cowards wrt trump and doge and usaid, to say nothing of how much anthropic le…especially when we have prominent rats and EAs being absolute cowards wrt trump and doge and usaid, to say nothing of how much anthrop…
↗Not briefed
hn
hn-ai
1moHyperFlow – one LoRA for all MiniMax-H3 tasks (ref2va/t2va/fl2va)—
↗Not briefed