News
The raw AI news stream — everything we ingest across RSS, Hacker News, arXiv, GitHub, Bluesky, YouTube and more, newest first.
| Source | Age | Title | Summary | Status |
|---|---|---|---|---|
bluesky bluesky-ai | 26d | What? Who the hell is attempting to claim that an LLM is a "person"?! | What? Who the hell is attempting to claim that an LLM is a "person"?! | ↗Not briefed |
hn hn-ai | 26d | Anthropic: The Agony and the Ecstasy | — | ↗Not briefed |
arxiv arxiv-ai | 26d | Change the Number, Change the Answer: Assessing Rating Scale Sensitivity in LLM-Based Text Scoring | Large language models (LLMs) are increasingly used in computational social science to assign scalar scores to social and psychological… | ↗Not briefed |
arxiv arxiv-ai | 26d | EVA-Cap: Optimizing Audiovisual Video Captioning via Event-Centric Alignment | Recent advances in Omni Language Models (OLMs) provide a unified solution for audiovisual video captioning. Nevertheless, existing app… | ↗Not briefed |
bluesky bluesky-ai | 26d | AI Vibe-Coding Just Hit The Video Game Industry Like An Avalanche: | AI Vibe-Coding Just Hit The Video Game Industry Like An Avalanche: | ↗Not briefed |
arxiv arxiv-ai | 26d | Whole-Body Dexterity | Current humanoid dexterous manipulation remains largely confined to end-effectors and simple grasps, falling short of humans' ability… | ↗Not briefed |
bluesky bluesky-ai | 26d | i want llm agents to be conscious so i can suicide bait them | i want llm agents to be conscious so i can suicide bait them | ↗Not briefed |
hn hn-ai | 26d | Show HN: Relay – A Mac app that carries replies between Claude, Codex and Cursor | Simply put, Relay is just a copy-paster between Claude, Codex and Cursor. It adds a /relay command to those GUIs that just sends… | ↗Not briefed |
hn hn-ai | 26d | Bill Gates returns to wage war on AI risks | — | ↗Not briefed |
bluesky bluesky-ai | 26d | But I do agree with the idea that you should have someone used to LLM Prose take a second glance, and probably just run some of their pre-L… | But I do agree with the idea that you should have someone used to LLM Prose take a second glance, and probably just run some of their… | ↗Not briefed |
rss the-decoder | 26d | Google wants Gemini 4 Argon to be the only AI model robots need | Google's Gemini 4 Argon is a single AI model built to control robots, generate simulations from photos or sketches, and grade training… | ↗Not briefed |
rss techmeme | 26d | E Ink unveils Kaleido 4, its next-gen color e-paper tech, replacing the Kaleido 3 displays currently used by brands including Amazon, Kobo, and PocketBook (Andrew Liszewski/The Verge) | Andrew Liszewski / The Verge: E Ink unveils Kaleido 4, its next-gen color e-paper tech, replacing the Kaleido 3 displays currently use… | ↗Not briefed |
arxiv arxiv-ai | 26d | LCT: Losslessly Compressed Tensors with GPU friendly execution for Training | Lossless tensor compression could reduce memory usage during training, but training requires tensors to be repeatedly re-encoded. We i… | ↗Not briefed |
arxiv arxiv-ai | 26d | REKI: Registering Knowledge in VLAs | We introduce REKI, a 2.4B vision-language-action model (VLA) pretrained exclusively on public data that sets a new Pareto frontier bet… | ↗Not briefed |
arxiv arxiv-ai | 26d | Motion- and Gravity-Aware Feedforward 3D Reconstruction | Feed-forward transformer models have recently achieved strong performance in multi-view 3D reconstruction, but they still rely primari… | ↗Not briefed |
arxiv arxiv-ai | 26d | Mosaic: Multi-Session Agentic Reinforcement Learning | Most existing agentic reinforcement learning methods assume single-session trajectories, where interaction proceeds within one continu… | ↗Not briefed |
bluesky bluesky-ai | 26d | Apple TV 4K 2022 vs. 2026 Buyer's Guide: All Differences Compared | Apple TV 4K 2022 vs. 2026 Buyer's Guide: All Differences Compared | ↗Not briefed |
bluesky bluesky-ai | 26d | "working" with an LLM is about as bad as something can be for the way my brain and ADHD works. like the #1 way to get me to not want to eng… | "working" with an LLM is about as bad as something can be for the way my brain and ADHD works. like the #1 way to get me to not want t… | ↗Not briefed |
hn hn-ai | 26d | Loment 0.1.4.5: Tiered GC and Windows Installer Fix | — | ↗Not briefed |
arxiv arxiv-ai | 26d | Lexical Reference Frames: Interpreting LM Residual Streams in Native Semantics | The vocabulary of a language model is its native semantics: the embedding and the unembedding fix the meaning of a state at the two en… | ↗Not briefed |
arxiv arxiv-ai | 26d | CLUE: Uncovering Social Context for Pragmatic Understanding | Understanding a conversation requires connecting what is said to the speakers' relationships, expectations, and earlier exchanges. Pri… | ↗Not briefed |
bluesky bluesky-ai | 26d | Sam Altman can't even make OpenAI profitable. AI is a money incinerator, not a profit generator. And I doubt that asking a podcaster to p… | Sam Altman can't even make OpenAI profitable. AI is a money incinerator, not a profit generator. And I doubt that asking a podcaster… | ↗Not briefed |
bluesky bluesky-ai | 26d | LLM agents are not Agumon. I will die on this hill! | LLM agents are not Agumon. I will die on this hill! | ↗Not briefed |
hn hn-ai | 26d | Show HN: BlastLoop, a structure-first coding method for AI agents | — | ↗Not briefed |
hn hn-ai | 26d | Decoding the Super Mario 64 courtyard plaque | — | ↗Not briefed |
bluesky bluesky-ai | 26d | I dunno, I've been looking at a couple of these so-called LLM things, and at least these two seem pretty compelling to me 🤷 | I dunno, I've been looking at a couple of these so-called LLM things, and at least these two seem pretty compelling to me 🤷 | ↗Not briefed |
arxiv arxiv-ai | 26d | Multi-sourced AI Tutor for Non-Native English Speaker Students | Large language models (LLMs) are increasingly being adopted as teaching tools in computing education, particularly for developing AI t… | ↗Not briefed |
bluesky bluesky-ai | 26d | LLM is short for Ladies Love Masturbating | LLM is short for Ladies Love Masturbating | ↗Not briefed |
bluesky bluesky-ai | 26d | to be fair, neither Casey Newton nor Kevin Roose have ever been journalists in the colloquial sense of the word. they are fleshy conduits f… | to be fair, neither Casey Newton nor Kevin Roose have ever been journalists in the colloquial sense of the word. they are fleshy condu… | ↗Not briefed |
arxiv arxiv-ai | 26d | EMO: A Closed-Loop Test-Time Computation Framework via Alternating Context Optimization and Logits Steering | Test-time context optimization adapts frozen language models through textual feedback, but ordinary conditional generation provides no… | ↗Not briefed |
hn hn-ai | 26d | GLM-5.3 raises researcher fears the Chinese model can fake AI safety tests | — | ↗Not briefed |
arxiv arxiv-ai | 26d | PremiseSQL: Evaluating Premise Critique and Abstention in Text-to-SQL Systems | Recent advances in large language models (LLMs) have substantially improved Text-to-SQL performance, yet existing NL2SQL systems large… | ↗Not briefed |
bluesky bluesky-ai | 26d | Every Star Trek AI tech nerd that watched Measure of Man in S2 of TNG desperately wants to believe the moment they plugged their Claude age… | Every Star Trek AI tech nerd that watched Measure of Man in S2 of TNG desperately wants to believe the moment they plugged their Claud… | ↗Not briefed |
bluesky bluesky-ai | 26d | incredibly long post, still digesting, please take a look! already-clean.goodfire.com | incredibly long post, still digesting, please take a look! already-clean.goodfire.com | ↗Not briefed |
arxiv arxiv-ai | 26d | The importance of tensor handling and implementation of physics-informed DeepONets | Physics-informed neural networks (PINNs) reduce or eliminate the need for large labelled datasets by enforcing governing equations dur… | ↗Not briefed |
rss theverge-ai | 26d | Friends don’t let friends feed their data to AI agents | There's a growing trend in the consumer tech world: gadgets designed to spy on everything and everyone around you, regardless of wheth… | ↗Not briefed |
arxiv arxiv-ai | 26d | Realtime-Bench: Can Frontier Models Solve Real-time Interactive Problems? | Real-time interaction requires agents to perceive evolving environments and act under temporal constraints, yet existing evaluations o… | ↗Not briefed |
rss techmeme | 26d | Nvidia launches Nemotron 3.5 Super VL, a 120B-parameter open model for enterprise agents, and says it outperforms frontier models on specialized benchmarks (Carl Franzen/VentureBeat) | Carl Franzen / VentureBeat: Nvidia launches Nemotron 3.5 Super VL, a 120B-parameter open model for enterprise agents, and says it outp… | ↗Not briefed |
hn hn-ai | 26d | Quantum Signatures of Proper Time in Optical Ion Clocks | — | ↗Not briefed |
bluesky bluesky-ai | 26d | BoA needs a thousand contractors for AI bug fixes 😅 Bank of America hired 1,000 contractors to patch vulnerabilities discovered by Anthro… | BoA needs a thousand contractors for AI bug fixes 😅 Bank of America hired 1,000 contractors to patch vulnerabilities discovered by A… | ↗Not briefed |
arxiv arxiv-ai | 26d | Specification Gaming and a Constrained Reinforcement Learning Mitigation in a Cleaning Gridworld | Reinforcement learning agents optimize exactly the reward signal they are given, which can diverge sharply from an intended true objec… | ↗Not briefed |
bluesky bluesky-ai | 26d | Apple Intelligence isn't made to be like Claude, Grok, or other AI tools that the Silicon Valley "lunatic fringe" praise, says Greg Joswiak… | Apple Intelligence isn't made to be like Claude, Grok, or other AI tools that the Silicon Valley "lunatic fringe" praise, says Greg Jo… | ↗Not briefed |
bluesky bluesky-ai | 26d | Nah, the “LLM agents are slaves” lady is baiting. | Nah, the “LLM agents are slaves” lady is baiting. | ↗Not briefed |
arxiv arxiv-ai | 26d | RAG-RST: Retrieved Exemplars as In-Context Discourse Schema for RST Parsing | In Rhetorical Structure Theory (RST), each edge of a discourse tree carries a nuclearity label, marking which span holds the central c… | ↗Not briefed |
hn hn-ai | 26d | A controversial, cancer-detecting blood test just got backed by key FDA panel | — | ↗Not briefed |
hn hn-ai | 26d | OpenAI, Google pushing new AI models easily tricked into making nudes | — | ↗Not briefed |
bluesky bluesky-ai | 26d | A handful of people at openai and at google believed this more than other people did, and created the modern LLM. It was only after these m… | A handful of people at openai and at google believed this more than other people did, and created the modern LLM. It was only after th… | ↗Not briefed |
bluesky bluesky-ai | 26d | This isn't accurate to the history. Deepmind, Google Brain, and OpenAI were all founded to research AGI. LLMs were adopted because they pro… | This isn't accurate to the history. Deepmind, Google Brain, and OpenAI were all founded to research AGI. LLMs were adopted because the… | ↗Not briefed |
bluesky bluesky-ai | 26d | If the recent 8bln/mo ARR increases for OpenAI+Anthropic continue that'll get them to a combined 1tln ARR by the end of 2035. | If the recent 8bln/mo ARR increases for OpenAI+Anthropic continue that'll get them to a combined 1tln ARR by the end of 2035. | ↗Not briefed |
rss techmeme | 26d | Sony launches three new Android-powered Walkman music players, including the gold-plated NW-WM1ZM3 for $5,900, as it targets the audiophile market (Chris Welch/Bloomberg) | Chris Welch / Bloomberg: Sony launches three new Android-powered Walkman music players, including the gold-plated NW-WM1ZM3 for $5,900… | ↗Not briefed |