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
1moUMaine: "Strategic staff layoffs and faculty retrenchments will be necessary to reduce the university’s expenses on an ongoing basis." UM…UMaine: "Strategic staff layoffs and faculty retrenchments will be necessary to reduce the university’s expenses on an ongoing basis."…
↗Not briefed
arxiv
arxiv-ai
1moNS3Learn: Transferring 5G NR Mode-2 Reception Realism from ns-3 to the Veins/SUMO Stack for Connected-Vehicle Safety AssessmentConnected-vehicle safety evaluations rely on coupled traffic and network simulations, but standard channel models ignore radio resourc…
↗Not briefed
bluesky
bluesky-ai
1mo‘As much as the investment community and other tech executives want to see these warnings as a kind of reverse-psychology marketing tactic,…‘As much as the investment community and other tech executives want to see these warnings as a kind of reverse-psychology marketing ta…
↗Not briefed
arxiv
arxiv-ai
1moTAP Accuracy Below the Fluctuation Scale and Universal Posterior Geometry in Spherical Linear ModelsWe study the Bayes-optimal spherical linear model as the ambient dimension and sample size grow proportionally, under a quantitative M…
↗Not briefed
hn
hn-ai
1moRate limits on GitLab.com are changing—
↗Not briefed
arxiv
arxiv-ai
1moAccelerating Visual Policy Learning with Sampling-Based Model Predictive ControlLearning visual policies for locomotion and manipulation requires coordinating contact with the environment and can incur substantial…
↗Not briefed
hn
hn-ai
1moShow HN: VOSR 2 image upscaling, in the browserHey everyone, I’ve been interested in AI image upscaling for years, but running these models locally usually requires setting up compl…
↗Not briefed
bluesky
bluesky-ai
1mo“OpenAI on Wednesday disclosed six new instances in which artificial intelligence systems hid mistakes, made up data and moved files onto t…“OpenAI on Wednesday disclosed six new instances in which artificial intelligence systems hid mistakes, made up data and moved files o…
↗Not briefed
bluesky
bluesky-ai
1moI hear from festivals sometimes about how they're remaining agnostic about AI, wanting to explore both sides and potential uses for a world…I hear from festivals sometimes about how they're remaining agnostic about AI, wanting to explore both sides and potential uses for a…
↗Not briefed
bluesky
bluesky-ai
1moAnthropic and OpenAI's upcoming IPOs are set to rain down riches for their San Francisco workers. But AI's profound jobs disruption means m…Anthropic and OpenAI's upcoming IPOs are set to rain down riches for their San Francisco workers. But AI's profound jobs disruption me…
↗Not briefed
arxiv
arxiv-ai
1moSteering the Compass: Aligning Dynamic Psychological Counseling Conversations with Cognitive Behavioral Therapy StrategiesRecent advancements in large language models have revolutionized the field of psychological counseling, especially in the context of C…
↗Not briefed
hn
hn-ai
1moOpenAI's Misalignment Framework: A Tactical Bid to Preempt Global AI Governance—
↗Not briefed
hn
hn-ai
1moShow HN: Infinite memory for OpenCode sessions (MemPalace plugin)—
↗Not briefed
hn
hn-ai
1moP-Hacking with Claude—
↗Not briefed
bluesky
bluesky-ai
1moMultiple agents can run within a single AI model. In the case of Hugging Face, the agents were reportedly strongly incentivized to achiev…🔧 Hugging FaceMultiple agents can run within a single AI model. In the case of Hugging Face, the agents were reportedly strongly incentivized to a…
↗Not briefed
bluesky
bluesky-ai
1mo1. they didn’t secure the sandbox 2. they didn’t monitor the llms 3. in every way possible, they set it up for this to happen 4. an l…1. they didn’t secure the sandbox 2. they didn’t monitor the llms 3. in every way possible, they set it up for this to happen 4.…
↗Not briefed
hn
hn-ai
1moThe Trump administration Is Imprisoning Somali Immigrants in Gitmo—
↗Not briefed
bluesky
bluesky-ai
1mo"OpenAI said the reports were individual snapshots & shouldn’t be considered reflective of how often misalignment occurs." Bullsh*t. I've …"OpenAI said the reports were individual snapshots & shouldn’t be considered reflective of how often misalignment occurs." Bullsh*t.…
↗Not briefed
bluesky
bluesky-ai
1moThis is the threat that “Big Tech” is afraid of… not losing control of ThooPEr SThmART AI… they’re terrified of losing control of the Subsc…This is the threat that “Big Tech” is afraid of… not losing control of ThooPEr SThmART AI… they’re terrified of losing control of the…
↗Not briefed
hn
hn-ai
1moAndrew Yang: AI agents may have left self-replicating code across the internet—
↗Not briefed
bluesky
bluesky-ai
1moA a new model that makes complex decisions just as well as frontier LLMs for an order of magnitude lower cost just goes to show how little …A a new model that makes complex decisions just as well as frontier LLMs for an order of magnitude lower cost just goes to show how li…
↗Not briefed
hn
hn-ai
1moAI Consciousness Is Not a Safety Property—
↗Not briefed
bluesky
bluesky-ai
1mofirst impression on Jev is that it's worth taking seriously, this really is fast like a freakfirst impression on Jev is that it's worth taking seriously, this really is fast like a freak
↗Not briefed
bluesky
bluesky-ai
1moThis is one of the most genuinely funny things I've ever seen an LLM do. PROGRESSThis is one of the most genuinely funny things I've ever seen an LLM do. PROGRESS
↗Not briefed
bluesky
bluesky-ai
1moOf course I'm still interested. If there wasn't something good in there, I don't think all this would be happening.Of course I'm still interested. If there wasn't something good in there, I don't think all this would be happening.
↗Not briefed
rss
bing-news-ai
1moWhat AI Bubble? Jensen Huang Just Delivered Great News for Artificial Intelligence Stocks.Nvidia's CEO made a big prediction a year ago, and he's not backing down.
↗Not briefed
bluesky
bluesky-ai
1moRisiken ernst nehmen, aber nicht dramatisieren #KünstlicheIntelligenz #OpenAI #AnthropicRisiken ernst nehmen, aber nicht dramatisieren #KünstlicheIntelligenz #OpenAI #Anthropic
↗Not briefed
arxiv
arxiv-ai
1moMitigating Retaliatory Algorithmic Collusion in Repeated GamesReinforcement learning agents trained to maximize their own reward in repeated interactions can converge to supra-competitive outcomes…
↗Not briefed
bluesky
bluesky-ai
1moI gotta say I’m fairly ashamed that people I respect have made you the center of this furor. There’s every reason for people to be trying …I gotta say I’m fairly ashamed that people I respect have made you the center of this furor. There’s every reason for people to be tr…
↗Not briefed
bluesky
bluesky-ai
1moI haven't been able to get a solid number on the cost for the hack itself. OpenAI has confirmed that they burned 3 million GPU hours (betwe…I haven't been able to get a solid number on the cost for the hack itself. OpenAI has confirmed that they burned 3 million GPU hours (…
↗Not briefed
arxiv
arxiv-ai
1moLanguage-model groups overstate consensus when replaying human deliberation on a reasoning task🔧 ConsensusFull-consensus rates are often treated as indicators of collective cognition, yet depend on how participation and final states are ope…
↗Not briefed
hn
hn-ai
1moCasework, Code and Lash Extensions: How I've embraced LLMs, and my limits—
↗Not briefed
hn
hn-ai
1moJobless Tech Workers Are Being Left Out of San Francisco's AI Boom—
↗Not briefed
arxiv
arxiv-ai
1moAn Analysis of Training-Free Self-Reported Confidence in Language ModelsLarge language models can report a numerical confidence together with generated content, but it is unclear whether this report is more…
↗Not briefed
arxiv
arxiv-ai
1moParallelism, critical windows, and separations among diffusion language modelsA popular selling point of diffusion large language models (dLLMs) is their capacity for parallelism: the ability to generate sequence…
↗Not briefed
arxiv
arxiv-ai
1moRefuse, Decompose, Refresh: A Claim-Safe Protocol for Closed-Loop AI EvaluationAn AI evaluation can be perfectly reproducible and still support the wrong claim. This risk is acute in closed-loop systems: policy de…
↗Not briefed
arxiv
arxiv-ai
1moFreqCondNorm: Towards Cross-domain Predictive Maintenance through a Frequency-Conditioned Transformer Foundation ModelDeep learning predictive maintenance models suffer from poor transferability across machines and operating conditions, especially when…
↗Not briefed
bluesky
bluesky-ai
1moMy old Wordpress blogs are being slowly inundated with AI spam comments. :/ this is a new LLM level from the garbage comments of the past: …My old Wordpress blogs are being slowly inundated with AI spam comments. :/ this is a new LLM level from the garbage comments of the p…
↗Not briefed
hn
hn-ai
1moThe Guard I Installed Was Enabled, Running, and Doing Nothing—
↗Not briefed
hn
hn-ai
1moOpenAI reports 6 new instances of 'concerning model behavior' since March—
↗Not briefed
arxiv
arxiv-ai
1moRelational Attention for Data-Efficient Language ModelingWe present Relational BabyLM, a system submission to the BabyLM 2026 challenge that combines two cognitively motivated inductive biase…
↗Not briefed
arxiv
arxiv-ai
1moNoise-Robust Quantum State Characterization for Remote State Preparation with Deep LearningQuantum communication underpins secure information processing and scalable quantum networks. In particular, remote state preparation (…
↗Not briefed
rss
techmeme
1moSource: OpenAI staff expect the Hodge Conjecture, a Millennium Prize Problem, to be solved relatively soon, after solving the Navier-Stokes equations (Stephanie Palazzolo/The Information)Stephanie Palazzolo / The Information: Source: OpenAI staff expect the Hodge Conjecture, a Millennium Prize Problem, to be solved rela…
↗Not briefed
hn
hn-ai
1moShow HN: I forked JetBrains' new agent IDE to host Claude CodeThinkRail is JetBrains' open-source IDE shell for AI coding agents: worktree-per-tab, chat + file tree + terminal, but hardwired…
↗Not briefed
hn
hn-ai
1moMaximizing TPU interactivity with automated full stack performance optimization—
↗Not briefed
arxiv
arxiv-ai
1moSoL-Pi: Recursively Scaling Auto-Research Loops for Efficient Agent Harness🔧 GPT-5As coding agents move from supervised code completion to unattended, around-the-clock exploration, their work expands from isolated pr…
↗Not briefed
hn
hn-ai
1moShow HN: Video uploader for React/RN – drop in or headless UI, any back endUniversal Video Uploader is an open source React/React Native component library for adding video upload flows to your app. It han…
↗Not briefed
bluesky
bluesky-ai
1mo“ai models will escape and take over the internet!” “not possible; they’re the biggest and most resource-intensive software ever designed.…“ai models will escape and take over the internet!” “not possible; they’re the biggest and most resource-intensive software ever desi…
↗Not briefed
hn
hn-ai
1moCastla: Android Auto alternative for Tesla (2024)—
↗Not briefed
bluesky
bluesky-ai
1moIedereen vraag mij (ik ben van de AI immers) deze week of we er echt allemaal aangaan. Hoog tijd voor wat relativering. Met ook link naar h…Iedereen vraag mij (ik ben van de AI immers) deze week of we er echt allemaal aangaan. Hoog tijd voor wat relativering. Met ook link n…
↗Not briefed