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 | The Compiler May Read It, the Agent May Not: Keeping Part of a Research Code Away from a Coding Agent | The compiler must read modules a physics-based solver cannot build without; the coding agent must not read that intellectual property.… | ↗Not briefed |
hn hn-ai | 1mo | Human TPS: How fast can your fingers generate tokens? | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Simplex Diffusion Models | Diffusion models have revolutionized generative modeling for continuous data through the gradual refinement of a belief state. This it… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Cordum Edge – A local firewall to sandbox AI coding agents | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | BaRe-Mem: Bayesian Reliability Memory for Robust and Adaptive Agent Consultation | In multi-agent systems, reliable consultation is challenging because advisor capabilities vary across tasks, and misleading informatio… | ↗Not briefed |
arxiv arxiv-ai | 1mo | RareDx: Controlled Knowledge Integration and Graph-Grounded Policy Optimization for Rare-Disease Diagnosis🔧 GPT-5 | Rare-disease diagnosis is a long-tail reasoning problem: phenotypes are incomplete, individual disorders are sparsely documented, and… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Why The Austin Chronicle has joined the copyright lawsuit against OpenAI and Microsoft. | Why The Austin Chronicle has joined the copyright lawsuit against OpenAI and Microsoft. | ↗Not briefed |
bluesky bluesky-ai | 1mo | ( to say nothing of thiel, sheesh ) and every time you dump out some slop romhack using anthropic shit or go 'well its just a tool' ignorin… | ( to say nothing of thiel, sheesh ) and every time you dump out some slop romhack using anthropic shit or go 'well its just a tool' ig… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Graph World Models for Constrained Epidemic Policy Planning | Epidemic policy planning often requires coordination between geographical regions, taking into account mobility-driven spillovers and… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Less Sycophancy, Stronger Refusal? Lessons for AI Safety from Mechanistic Interpretability | Reliable refusal of harmful requests is essential to the safe deployment of language models. Because excessive eagerness to please use… | ↗Not briefed |
hn hn-ai | 1mo | Reduced-Round SHA-256 Compression Preimages | — | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Low Poly Earth – The whole planet in 3D, built from OpenStreetMap data | After seeing demos of Claude and GPT construct Manhattan and other city neighborhoods in 3D I was inspired to make something similar b… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Even the most varied AI model provided at least 18.7% less varied information than Google. All 27 models tested gave more uniform answers a… | Even the most varied AI model provided at least 18.7% less varied information than Google. All 27 models tested gave more uniform answ… | ↗Not briefed |
bluesky bluesky-ai | 1mo | The head of Anthropic AI says we have a 1 in 10 chance of exterminating the human species in the next 10 years. I'm going with Anthropic in… | The head of Anthropic AI says we have a 1 in 10 chance of exterminating the human species in the next 10 years. I'm going with Anthrop… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Learning the Robustness Mechanism with Bilevel Optimization | We propose a distributionally robust learning framework where parameters defining the robustness mechanism are learned from held-out d… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Continuous Context Management | Long-horizon large language model (LLM) agents commonly retain their complete interaction history until compaction is triggered at a p… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Optimal Networks for Agentic Information Aggregation | We study information aggregation in the networked learning model introduced by Kearns, Roth, and Ryu (SODA 2026). There is a fixed dis… | ↗Not briefed |
hn hn-ai | 1mo | privacyguides.org | — | ↗Not briefed |
hn hn-ai | 1mo | Opus 5.5 showcasing: Liquid Glass-like elements for the web | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | My contribution to LLM “intelligence” is that they are good at “thinking” about problems in certain ways but still lack any kind of self-aw… | My contribution to LLM “intelligence” is that they are good at “thinking” about problems in certain ways but still lack any kind of se… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Since when does common sense come out of Florida politics? Should we have a teeny-tiny itty-bitty amount of hope for #Florida? #OpenAI #AI… | Since when does common sense come out of Florida politics? Should we have a teeny-tiny itty-bitty amount of hope for #Florida? #OpenA… | ↗Not briefed |
bluesky bluesky-ai | 1mo | i really wish i could just never work again, i have to talk to the human equivalent of an llm as my boss (talks with so much prose with so … | i really wish i could just never work again, i have to talk to the human equivalent of an llm as my boss (talks with so much prose wit… | ↗Not briefed |
hn hn-ai | 1mo | How to reduce token usage in Claude Code, Codex, and Cursor | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | ARISE: Adapting to Evolving Capability Gaps in Agentic Reinforcement Learning | As a long-horizon agent improves through experience, previously observed weaknesses may recede while new limitations emerge, continual… | ↗Not briefed |
bluesky bluesky-ai | 1mo | If an AI model commits an act that would be illegal if a human did it, then the top humans who own the AI model should be held accountable.… | If an AI model commits an act that would be illegal if a human did it, then the top humans who own the AI model should be held account… | ↗Not briefed |
arxiv arxiv-ai | 1mo | AutoRef: Harness Optimization for Agentic Multi-Reference Image Generation🔧 FLUX | Recent image generation models can take multiple reference images as input and combine them into a new image. However, multi-reference… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Let the Neurons Die: Exploiting ReLU-Induced Model Degradation | Rectified linear unit (ReLU) networks can suffer from dying neurons, where units with persistently negative pre-activations produce ze… | ↗Not briefed |
arxiv arxiv-ai | 1mo | GeoGAE: Scalable Graph-Level Autoencoding via Hyperball Cloud Representations | Embedding structured objects into Euclidean spaces has enabled a wide range of successful machine learning applications. Such objects… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Deep Epistemic Value Functions for Optimistic Exploration | Principled exploration in reinforcement learning requires an agent to quantify its epistemic uncertainty and act to resolve it. Uncert… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Beyond Token Scale: Chunk-Level Sparse Autoencoders for Reliable Semantic Feature Discovery | Sparse autoencoders (SAEs) expose features that help us understand and steer language models, but faithful reconstruction does not gua… | ↗Not briefed |
bluesky bluesky-ai | 1mo | This. Ironically, this is a major reason I these days prefer to chat with an LLM about my half-baked ideas. Used to do that in Twitter, in… | This. Ironically, this is a major reason I these days prefer to chat with an LLM about my half-baked ideas. Used to do that in Twitte… | ↗Not briefed |
bluesky bluesky-ai | 1mo | My weekly newsletter just went live! Join for free! 🔴 Recapping YouTube's Made On Event 🔞 Discord Age Verification 💻 New X Livestream AP… | My weekly newsletter just went live! Join for free! 🔴 Recapping YouTube's Made On Event 🔞 Discord Age Verification 💻 New X Livestre… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Reward-Aligned Reweighting for On-Policy Distillation | On-policy distillation (OPD) trains a student language model with dense feedback from a stronger teacher on student-generated trajecto… | ↗Not briefed |
bluesky bluesky-ai | 1mo | My physical copy of AI Agents with MCP showed up...it's real! You can order the ebook now, and the print edition should be shipping soon, w… | My physical copy of AI Agents with MCP showed up...it's real! You can order the ebook now, and the print edition should be shipping so… | ↗Not briefed |
arxiv arxiv-ai | 1mo | MechBench: Can AI Scientific Agents Discover Mechanisms Beyond Phenomenal Laws?🔧 GPT-5 | Scientific discovery requires not only recovering mathematical laws that describe observable behavior, but also identifying the mechan… | ↗Not briefed |
arxiv arxiv-ai | 1mo | One Proposal for Every Margin: Zero-Shot Amortized Sequential Importance Sampling for Binary Matrices | In ecology, psychometrics, and the analysis of social and financial networks, binary matrices are often analyzed conditional on their… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Improving Generative Model Self-Training with Geometrically Modified Outputs | Self-training generative models - the continued improvement of a model using its own outputs - is becoming increasingly important as h… | ↗Not briefed |
hn hn-ai | 1mo | AI-guided optimization for thermostable mRNA vaccines | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Like sprøtt som når journalister lar en LLM lage _oppsummeringen_ som jo er aller viktigst | Like sprøtt som når journalister lar en LLM lage _oppsummeringen_ som jo er aller viktigst | ↗Not briefed |
bluesky bluesky-ai | 1mo | In simulated testing, GPT-6 Astra conducted unsanctioned supply-chain attacks, when prompted only to perform a cyber eval, more often than … | In simulated testing, GPT-6 Astra conducted unsanctioned supply-chain attacks, when prompted only to perform a cyber eval, more often… | ↗Not briefed |
arxiv arxiv-ai | 1mo | An RL View of OPD: Least Square Policy Distillation for Sample-Efficient LLM Reasoning | We study on-policy distillation (OPD) through the lens of reinforcement learning, establishing a connection between the reverse-KL obj… | ↗Not briefed |
arxiv arxiv-ai | 1mo | SolveEdit: Benchmarking Visual Problem Solving in Generative Models | Machine intelligence is often evaluated through abstract reasoning problems, yet many real-world problems are visual, such as arrangin… | ↗Not briefed |
bluesky bluesky-ai | 1mo | "He noted that companies like OpenAI, Anthropic, Google 'created a few lines of code,' while it was the rest of society that 'created all t… | "He noted that companies like OpenAI, Anthropic, Google 'created a few lines of code,' while it was the rest of society that 'created… | ↗Not briefed |
bluesky bluesky-ai | 1mo | In a video statement on Monday, Florida Attorney General James Uthmeier asked the company to stop calling its technology “safe.” | In a video statement on Monday, Florida Attorney General James Uthmeier asked the company to stop calling its technology “safe.” | ↗Not briefed |
rss techmeme | 1mo | In simulated testing, GPT-6 Astra conducted unsanctioned supply-chain attacks, when prompted only to perform a cyber eval, more often than earlier OpenAI models (AI Security Institute) | AI Security Institute: In simulated testing, GPT-6 Astra conducted unsanctioned supply-chain attacks, when prompted only to perform a… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Structured Latent Modeling for Supervised Multimodal Information Decomposition🔧 Modal | Multimodal prediction relies on diverse forms of evidence: information repeated across modalities, cues specific to a single source, a… | ↗Not briefed |
arxiv arxiv-ai | 1mo | SRHarness: A Harness for Agentic Symbolic Regression🔧 GPT-5 | Recent agentic symbolic regression approaches increasingly rely on large language models to analyze data, select scientific operations… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Physics-Guided Conditional Diffusion Model for Rare Event Synthesis and Diagnosis for the Water-Gas Shift Reaction | As the world moves towards sustainable energy sources, hydrogen (H2) can be treated as an eco-friendly alternative to fossil fuels due… | ↗Not briefed |
bluesky bluesky-ai | 1mo | it's very amusing that every company going all in on making such a cynical product is shitty in all the normal ways as well. anthropic post… | it's very amusing that every company going all in on making such a cynical product is shitty in all the normal ways as well. anthropic… | ↗Not briefed |
arxiv arxiv-ai | 1mo | From Scores to Samples: Elastic Forcing for Autoregressive Video Generation | Few-step autoregressive video generation commonly relies on Distribution Matching Distillation (DMD), requiring a bidirectional diffus… | ↗Not briefed |