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
16d*waiting for the AI partisans to tell me LLM maps are better than actual ones*waiting for the AI partisans to tell me LLM maps are better than actual ones
↗Not briefed
bluesky
bluesky-ai
16dI have once again defeated a LLM in the battle of witsI have once again defeated a LLM in the battle of wits
↗Not briefed
hn
hn-ai
16dQuasi-Riemann, Biologic Discovery, and Swanson's "Undiscovered Public Knowledge"—
↗Not briefed
bluesky
bluesky-ai
16dword on the street is that there is *another* PHP vulnerability. this one was found by something even more powerful for security research …word on the street is that there is *another* PHP vulnerability. this one was found by something even more powerful for security rese…
↗Not briefed
bluesky
bluesky-ai
16dAnthropic: We don't have data to show that pushing our AI tools into schools had a positive impact www-cdn.anthropic.com/a9ceec76a54d...Anthropic: We don't have data to show that pushing our AI tools into schools had a positive impact www-cdn.anthropic.com/a9ceec76a5…
↗Not briefed
arxiv
arxiv-ai
16dTerraTransfer++: Demonstration-Free End-to-End PlannerEnd-to-end autonomous driving has achieved state-of-the-art performance on benchmarks and real-world deployments. Its standard trainin…
↗Not briefed
hn
hn-ai
16dAnthropic's Public Benefit Charter—
↗Not briefed
bluesky
bluesky-ai
16dThis is just psychopath behavior to me. Treating humans as expendable, making a personal calculation to pick and choose other people’s live…This is just psychopath behavior to me. Treating humans as expendable, making a personal calculation to pick and choose other people’s…
↗Not briefed
hn
hn-ai
16dWhat I saw inside Anthropic: AI lab pushes towards recursive self-improvement—
↗Not briefed
bluesky
bluesky-ai
16dHere's a preliminary look at track verification for Super Typhoon #Champi in the West Pacific. I selected a variety of models, but the ones…Here's a preliminary look at track verification for Super Typhoon #Champi in the West Pacific. I selected a variety of models, but the…
↗Not briefed
hn
hn-ai
16dRun many agents 3x faster on your Mac—
↗Not briefed
bluesky
bluesky-ai
16dIf I construct an academic argument using another person's writing, I'm obliged to disclose and properly cite it. Why should machine-assist…If I construct an academic argument using another person's writing, I'm obliged to disclose and properly cite it. Why should machine-a…
↗Not briefed
arxiv
arxiv-ai
16dSciFusion: Physics-Guided Compositional Diffusion for Multimodal 3D Scientific Image GenerationExisting 3D generative approaches are typically developed for individual imaging modalities, requiring substantial modality-specific d…
↗Not briefed
bluesky
bluesky-ai
16dMammoth-1 27B parameters #decisions #model by @basetenco.bsky.social. Fine tuned for typed answers on #Qwen 3.5. #AI #LLM #SLM #JEV #OpenSo…🔧 Qwen 3Mammoth-1 27B parameters #decisions #model by @basetenco.bsky.social. Fine tuned for typed answers on #Qwen 3.5. #AI #LLM #SLM #JEV #O…
↗Not briefed
rss
huggingface-blog
16dNVIDIA Kumo Relational Outperforms Supervised Machine Learning on Relational Data, No Training Required—
↗Not briefed
hn
hn-ai
16dArchestra 1.4: OpenAPPA and Self Hosted Runtime for Claude Code and Codex—
↗Not briefed
hn
hn-ai
16dTrajecktory, an open-source, local-first job search dashboard—
↗Not briefed
bluesky
bluesky-ai
16dOh, I missed this earlier: "the underlying model versions are not disclosed to the author". I cannot imagine submitting a paper that said …Oh, I missed this earlier: "the underlying model versions are not disclosed to the author". I cannot imagine submitting a paper that…
↗Not briefed
hn
hn-ai
16dAnthropic posts $80B loss in first half of 2026—
↗Not briefed
bluesky
bluesky-ai
16dokay valve here’s my new recommendation: forget marking ai slop, instead have a system whereby people can submit their game to be marked as…okay valve here’s my new recommendation: forget marking ai slop, instead have a system whereby people can submit their game to be mark…
↗Not briefed
hn
hn-ai
16dSilkStack – Local ComfyUI image browser with on-device semantic search🔧 ComfyUI—
↗Not briefed
bluesky
bluesky-ai
16dElectricity was dangerous before it was convenient. Before it could be widely adopted, we had to find a way for electricity to fail small. …Electricity was dangerous before it was convenient. Before it could be widely adopted, we had to find a way for electricity to fail sm…
↗Not briefed
hn
hn-ai
16dCorridorKey 2.0: Neural green and blue screen keyer—
↗Not briefed
bluesky
bluesky-ai
16dseriously this is the corniest shit ever, and is a pretty gross slap in the face to the security and software communities from openai.seriously this is the corniest shit ever, and is a pretty gross slap in the face to the security and software communities from openai.
↗Not briefed
arxiv
arxiv-ai
16dGender, Sexual Orientation, and the One-Year Refugee Bar: A Computational Study of Bill C-12 and Canadian Refugee AdjudicationIn March 2026, Bill C-12 amended the Immigration and Refugee Protection Act to make refugee claims ineligible for referral to the Refu…
↗Not briefed
arxiv
arxiv-ai
16dMeasuring Research Progress and Linguistic Inclusion in Low-Resource LanguagesLarge Language Models (LLMs) have accelerated research on low-resource and multilingual NLP, but their effectiveness in creating an eq…
↗Not briefed
bluesky
bluesky-ai
16d"we need to have a tool that's not built around APIs, webhooks, multi-app integrations, AI, agentic automation, etc just a simple tool tha…"we need to have a tool that's not built around APIs, webhooks, multi-app integrations, AI, agentic automation, etc just a simple too…
↗Not briefed
bluesky
bluesky-ai
16dWith the information we have now I’m inclined to agree that OpenAI at least attempted to responsibly go about this coordinated disclosure …With the information we have now I’m inclined to agree that OpenAI at least attempted to responsibly go about this coordinated disclos…
↗Not briefed
bluesky
bluesky-ai
16dbased on this morning's experience running non-stock offensive-cybersecurity-capable Gemini Argon against various repos my conclusion is th…based on this morning's experience running non-stock offensive-cybersecurity-capable Gemini Argon against various repos my conclusion…
↗Not briefed
arxiv
arxiv-ai
16dEgoHand: Monocular 3D Hand Reconstruction from Egocentric VideoWe introduce EgoHand, a monocular 3D hand reconstruction model that achieves state-of-the-art performance on egocentric video. Central…
↗Not briefed
arxiv
arxiv-ai
16dGAV: A scientific experimentation environment for PDE agentsReliable feedback is essential for coding agents to develop programs for partial differential equations (PDEs). Yet evaluating these p…
↗Not briefed
arxiv
arxiv-ai
16dGeneralizing Metropolis-Hastings for DAG SamplingA fundamental challenge in Bayesian network structure learning lies in efficiently approximating the posterior distribution of directe…
↗Not briefed
bluesky
bluesky-ai
16dICYMI: buttondown.com/surekhadavie...ICYMI: buttondown.com/surekhadavie...
↗Not briefed
bluesky
bluesky-ai
16dPHP armageddon: OpenAI, Google’s Wiz find ancient holes, panicking the internetPHP armageddon: OpenAI, Google’s Wiz find ancient holes, panicking the internet
↗Not briefed
arxiv
arxiv-ai
16dSWE-VERIFIER: Selective Verification for Cost-Efficient Coding AgentsCoding agents spend substantial computation generating actions and processing long interaction histories. Poor action choices can stee…
↗Not briefed
hn
hn-ai
16dLamentation on AI in Mathematical Research—
↗Not briefed
bluesky
bluesky-ai
16d40B parameter #coding #LLM #model #OpenSource Apache 2.0 #AI #SelfHosting 🤗 huggingface.co/RecursiveSI/...40B parameter #coding #LLM #model #OpenSource Apache 2.0 #AI #SelfHosting 🤗 huggingface.co/RecursiveSI/...
↗Not briefed
arxiv
arxiv-ai
16dA Quadratic Lens on Muon's Polar Update: Benefits, Pitfalls, and the Role of Newton-SchulzMuon updates matrix-valued parameters along approximately orthogonalized directions and often trains large language models faster than…
↗Not briefed
arxiv
arxiv-ai
16dBreaking the high-degree barrier in Lehmer's problem with a human-AI discovery loopWhile artificial intelligence has made rapid progress in mathematical reasoning, many open problems depend on structure that is reveal…
↗Not briefed
bluesky
bluesky-ai
16dopenai has a skill to check for the PHP 0day github.com/openai/php2s...openai has a skill to check for the PHP 0day github.com/openai/php2s...
↗Not briefed
arxiv
arxiv-ai
16dEdit, Don't Imitate: Minimal-Edit On-Policy Distillation for Multimodal ReasoningVisual mathematical reasoning can fail before reasoning begins: when a vision-language model (VLM) misreads a diagram, subsequent reas…
↗Not briefed
arxiv
arxiv-ai
16dLongZero: Enhancing Long-Context Reasoning in LLMs from Zero DataReinforcement learning (RL) is an effective way to strengthen the long-context reasoning of large language models (LLMs), but it needs…
↗Not briefed
bluesky
bluesky-ai
16dso uh from what that link says, openai was essentially doing extremely responsible disclosure and not at all dropping 0days on stage? that …so uh from what that link says, openai was essentially doing extremely responsible disclosure and not at all dropping 0days on stage?…
↗Not briefed
rss
huggingface-blog
16dTrain in Simulation, Deploy on a Real Robot: Microduck with Isaac Lab 3.0—
↗Not briefed
hn
hn-ai
16dShow HN: RMBG – remove backgrounds from images in the browser (no uploads)I am a person who does editing and digital designing. I used to use this tool, remove.bg, for getting transparent PNGs from either sto…
↗Not briefed
bluesky
bluesky-ai
16dInternet is reeeling from a php vulnerability being announced to everyone via malware developed by openai and it's scary enough for XIV to …Internet is reeeling from a php vulnerability being announced to everyone via malware developed by openai and it's scary enough for XI…
↗Not briefed
arxiv
arxiv-ai
16dThink Harder Where It Matters: Item-Level Adaptive Recursion for Sequential RecommendationExisting sequential recommenders usually allocate a fixed computational budget to every item in a user's interaction history, regardle…
↗Not briefed
arxiv
arxiv-ai
16dPhysiCalWorld: In-Context Physics Calibration for World ModelsExisting world models typically perceive a scene through current visual frame or a narrow history window. However, these observations…
↗Not briefed
arxiv
arxiv-ai
16dA Quasipolynomial Algorithm for $\ell_\infty$ ContractionsWe give an algorithm for computing an $\eps$-approximate fixed point of any $d$-dimensional nonexpansion with respect to the $\ell_\in…
↗Not briefed
arxiv
arxiv-ai
16dSelf-normalized Bernstein Inequalities for Regenerative Processes and Applications to Learning Without IndependenceWe prove a self-normalized Bernstein inequality for bounded functions of a regenerative process with independent, identically distribu…
↗Not briefed