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 |
|---|---|---|---|---|
arxiv arxiv-ai | 1mo | Interpolated Policy Distillation: A Controllable Continuum Between Off-Policy and On-Policy Distillation | Off-policy and on-policy distillation have traditionally been formulated as separate paradigms, each favoring a different property of… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Trajectory Soup: Pushing the Compute-Scaling Frontier of LLM Mid-training via Diverse Trajectories | Mid-training equips pretrained large language models with specialized and reasoning capabilities, but the returns of this stage are bo… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Disentangling Spurious Correlations in Vision-Language-Action Models via Predicting Domain-Invariant Latent Lookahead | Vision-Language-Action (VLA) models remain brittle under visual distribution shifts, often relying on spurious correlations tied to do… | ↗Not briefed |
bluesky bluesky-ai | 1mo | "Eine besondere Stärke von #OpenAI-Agenten ist, dass komplexe Workflows #autonom geplant werden." Insider-Tipp an die deutschen Medien: Es… | "Eine besondere Stärke von #OpenAI-Agenten ist, dass komplexe Workflows #autonom geplant werden." Insider-Tipp an die deutschen Medie… | ↗Not briefed |
bluesky bluesky-ai | 1mo | LLMってプログラマーの三大美徳（怠惰、短気、傲慢）の価値観の究極的な発露として出てきて、三大美徳を誰もが簡単に実現できるように発達したようなところがあるよね。 そしてその結果として、三大美徳を言う人が暗に含意してた「三大美徳を追求することで最大の生産性が発揮できるほど優秀に… | LLMってプログラマーの三大美徳（怠惰、短気、傲慢）の価値観の究極的な発露として出てきて、三大美徳を誰もが簡単に実現できるように発達したようなところがあるよね。 そしてその結果として、三大美徳を言う人が暗に含意してた「三大美徳を追求することで最大の生産性が発揮できる… | ↗Not briefed |
arxiv arxiv-ai | 1mo | The Vote Hides the Failure: Aggregation Choice and Noise Robustness in Heart Murmur Detection | Noise robustness in automated phonocardiogram (PCG) murmur detection, and how it is measured, remains underexamined despite growing in… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Guardrails for Claude Code: blocks rm -RF, reports what loaded | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Sources: Anthropic dedicated nearly a third of its IPO prospectus to detailing "risk factors", including that its AI may pose "existential … | Sources: Anthropic dedicated nearly a third of its IPO prospectus to detailing "risk factors", including that its AI may pose "existen… | ↗Not briefed |
rss techmeme | 1mo | Sources: Anthropic dedicated nearly a third of its IPO prospectus to detailing "risk factors", including that its AI may pose "existential risks to humanity" (Financial Times) | Financial Times: Sources: Anthropic dedicated nearly a third of its IPO prospectus to detailing “risk factors”, including that its AI… | ↗Not briefed |
arxiv arxiv-ai | 1mo | GLASS: Global Latent Aggregation with Slot-based Set Decoding for Scalable All-Atom Crystal Generation | Generative models for crystals enable the discovery of novel structures, but scaling all-atom generation to larger systems such as met… | ↗Not briefed |
arxiv arxiv-ai | 1mo | From Learner Behavior to Reusable Skills for Effective and Efficient Learner Simulation | Learner simulation aims to reproduce how a particular learner behaves on new tasks. Although Large Language Models (LLMs) can generate… | ↗Not briefed |
bluesky bluesky-ai | 1mo | 漏洩した個人情報、これまではダークウェブ公開が定番だったけど、かわりにLLM学習データに回収される未来が来たら笑う。(笑えない | 漏洩した個人情報、これまではダークウェブ公開が定番だったけど、かわりにLLM学習データに回収される未来が来たら笑う。(笑えない | ↗Not briefed |
arxiv arxiv-ai | 1mo | Lucid Dreaming for World Models: Learning to Doubt Imagination and Decide by Trust | World models enable agents to learn and plan in imagination, but predictions beyond their experience can become unreliable and mislead… | ↗Not briefed |
bluesky bluesky-ai | 1mo | "If such a breakthrough happened, the authors say, AI systems could rapidly eclipse human experts in most fields and “radically accelerate”… | "If such a breakthrough happened, the authors say, AI systems could rapidly eclipse human experts in most fields and “radically accele… | ↗Not briefed |
arxiv arxiv-ai | 1mo | When Tools Silently Lie: Evaluating and Mitigating Blind Compliance in Tool-Augmented Data Agents | Tool-augmented data agents rely on tool outputs for analytical decisions. Yet successful execution can return plausible but incorrect… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Just learned an amazing neologism to describe people who just parrot LLM outputs: meat proxies | Just learned an amazing neologism to describe people who just parrot LLM outputs: meat proxies | ↗Not briefed |
hn hn-ai | 1mo | Stanford and Nvidia's open Jev-like model | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | In a sane world, OpenAI would be shitting absolute bricks right now. This should be a High Court case. | In a sane world, OpenAI would be shitting absolute bricks right now. This should be a High Court case. | ↗Not briefed |
arxiv arxiv-ai | 1mo | Multimodal Detection of Higher-Order Behavioral Constructs: Self-Compassion in Structured Reflective Interaction | Many of the qualities that matter most in how people learn and grow, how someone regulates their emotions, reflects on a setback, or s… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Improved Distributional Diffusion Models | Distributional Diffusion Models (DDMs) replace the standard mean-prediction denoiser with a \emph{distributional} denoiser trained via… | ↗Not briefed |
arxiv arxiv-ai | 1mo | From Judgment Quality to Downstream Utility: Rethinking LLM-as-a-Judge for Open-Ended Tasks | LLM-as-a-Judge is increasingly used to evaluate policy responses on open-ended tasks that lack ground-truth answers. Existing work oft… | ↗Not briefed |
arxiv arxiv-ai | 1mo | LoLBench: Evaluating Coding Agents with Long-Horizon Proposals on Large Software Systems | Modern coding agents can deliver increasingly large repository-level changes, and recent benchmarks reflect this by emphasizing long-h… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Claude Sonnet 5.5 is one day old and already one-shot this entire 3D zombie FPS in 49 minutes for $177 #ai #artificialintelligence #sonnet… | Claude Sonnet 5.5 is one day old and already one-shot this entire 3D zombie FPS in 49 minutes for $177 #ai #artificialintelligence #s… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Here is the email obtained by the ABC that was sent by OpenAI to the Australian government's Services Australia to disclose the incident | Here is the email obtained by the ABC that was sent by OpenAI to the Australian government's Services Australia to disclose the incide… | ↗Not briefed |
hn hn-ai | 1mo | Getting AI 'drunk' makes it more likely to break rules and share secrets | — | ↗Not briefed |
rss techmeme | 1mo | Google appeals the European Commission's DMA mandates to open up Android at the EU General Court, arguing the requirements threaten user privacy and security (Samuel Stolton/Bloomberg) | Samuel Stolton / Bloomberg: Google appeals the European Commission's DMA mandates to open up Android at the EU General Court, arguing… | ↗Not briefed |
bluesky bluesky-ai | 1mo | So the Anthropic investors’ plan basically is: 1) make as much money as possible, 2) exit quickly, 3) build a bunker in time, then 4) …🔧 ChatGPT | So the Anthropic investors’ plan basically is: 1) make as much money as possible, 2) exit quickly, 3) build a bunker in time, then… | ↗Not briefed |
arxiv arxiv-ai | 1mo | SQUARE: Structured Quantum Representation Adapters as Compact Quadratic Feature Maps for Frozen Language Models | Frozen language models (LMs) are increasingly used as fixed feature extractors for downstream reranking, scoring, and preference model… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Train Ahead, Distill Back: Bootstrapping On-Policy Self-Distillation for Large Language Models | On-policy self-distillation (OPSD) improves large language models by letting a self-teacher with privileged information provide dense… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Overmux - Build custom agent dev environments with tmux and web UIs | I spend most of my time coding with agents, often from my phone. Off-the-shelf tools like Herdr and Orca look cool, but I didn’t want… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Seen someone claim that AI, in the context of art, is just another tool for expressing ideas. Ok if we goin with that, I'ma say that no mat… | Seen someone claim that AI, in the context of art, is just another tool for expressing ideas. Ok if we goin with that, I'ma say that n… | ↗Not briefed |
arxiv arxiv-ai | 1mo | SkillCome: Group Contrast Skill Optimization with Dual Memory | Skill evolution improves the capabilities of large language models by analyzing trajectories generated under a given skill and modifyi… | ↗Not briefed |
arxiv arxiv-ai | 1mo | LLM unbranding: Erasing Commercial Identity while Preserving Generic Utility | Establishing unbranding as a critical practice to prevent visual logos from acquiring negative connotations is standard in image gener… | ↗Not briefed |
arxiv arxiv-ai | 1mo | When Should Agents Check External State? Budgeting Observations for Stored Intentions | Prospective memory allows an agent to retain an intention tied to a future condition, but the stored intention does not reveal whether… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Learning the Structure of Triangular Transport Maps | Triangular transport maps provide a flexible approach to sampling-based probabilistic modeling, including density estimation, generati… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Cross-Linguistic Effects in Bilingual Phoneme BabyLMs | Cross-linguistic effects are a central topic in bilingual first-language acquisition. Artificial learners can help investigate L1-L2 i… | ↗Not briefed |
bluesky bluesky-ai | 1mo | well as long as they show a path to profitability in the near future- (checks notes) ok as gullible as wall street guys can be i don't th… | well as long as they show a path to profitability in the near future- (checks notes) ok as gullible as wall street guys can be i don… | ↗Not briefed |
bluesky bluesky-ai | 1mo | ⚠️ OpenAI will not release its newest artificial intelligence model, known as Astra 6.1, after internal testing by the ChatGPT-maker reveal…🔧 ChatGPT | ⚠️ OpenAI will not release its newest artificial intelligence model, known as Astra 6.1, after internal testing by the ChatGPT-maker r… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Dann löst euch auf?! | Dann löst euch auf?! | ↗Not briefed |
arxiv arxiv-ai | 1mo | Unlocking the Critic: Reward-Free Policy Optimization for LLM Post-Training | Recent approaches to reinforcement learning (RL) post-training for large language models increasingly remove the critic to reduce trai… | ↗Not briefed |
bluesky bluesky-ai | 1mo | People now insisting that I change MY language because it sounds like an LLM 😂 | People now insisting that I change MY language because it sounds like an LLM 😂 | ↗Not briefed |
arxiv arxiv-ai | 1mo | Multichannel Audio Quality Assessment: Extending Pretrained Perceptual Models to Spatial Audio | Accurate perceptual quality assessment is essential for evaluating and optimizing spatial audio, where perceived quality depends on bo… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Interpretable intrinsic dimension estimation through componentwise calibration of distance and angle | DANCo (Dimensionality from Angle and Norm Concentration) jointly calibrates nearest-neighbor distance and angular statistics and consi… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Neural Constitutive Learning for Generalized Reaction-Diffusion Systems | Generalized reaction-diffusion systems encompass diverse transport mechanisms and coupled reaction kinetics. A central question for ne… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Learning from Viable Failure Prefixes: Milestone Viability Potential Policy Optimization for Long-Horizon LLM Agents | Long-horizon LLM agents require reinforcement learning methods that can assign credit to intermediate decisions under sparse and delay… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Designing a Boundary Negotiating Artifact for Collaborative Socio-Technical Sense-Making in AI Regulatory Sandboxes | The rapid, unpredictable advancements in AI system capabilities has seen regulators take adaptive and experimental approaches to polic… | ↗Not briefed |
bluesky bluesky-ai | 1mo | www.theguardian.com/technology/2... | www.theguardian.com/technology/2... | ↗Not briefed |
bluesky bluesky-ai | 1mo | Editorial: Sam Altman is the son of a dermatologist and a real estate broker, which is why he makes your skin crawl. trib.al/S6Rc1Xn | Editorial: Sam Altman is the son of a dermatologist and a real estate broker, which is why he makes your skin crawl. trib.al/S6Rc1Xn | ↗Not briefed |
arxiv arxiv-ai | 1mo | VACE: Validation-Gated Alternating Co-Evolution of Agent Models and Harnesses | Language model agents can be improved by updating their model weights or refining the harness that guides task execution. These compon… | ↗Not briefed |
arxiv arxiv-ai | 1mo | What Does Post-Training Change in Multilingual Reasoning? | Open-source reasoning models provide unequal access to reasoning capability across languages. When a model can solve a problem but can… | ↗Not briefed |