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 |
|---|---|---|---|---|
hn hn-ai | 1mo | Show HN: AFAW – A multi-agent coding workflow that doesn't trust the model | — | ↗Not briefed |
hn hn-ai | 1mo | Composites for Stage 2 | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | T-ARC: Topology-Aware Randomized Clustering via Distributionally Robust Stochastic Block Models | In this work, we introduce a new clustering method, namely T-ARC (Topology-Aware Randomized Clustering), that corrects the geometric b… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Understanding Head Geometry and Dynamics in Federated Regression through a Natural Solution Selection Rule: An Unconstrained Feature Model Analysis | In federated averaging, local objectives can admit multiple optimal heads, making the aggregate depend on which heads clients return.… | ↗Not briefed |
rss openai-news | 1mo | Disrupting a coordinated model-distillation campaign | Learn how OpenAI disrupted a campaign to extract protected model reasoning and is strengthening defenses against adversarial distillat… | ↗Not briefed |
bluesky bluesky-ai | 1mo | llm use was also critical to my ability to 1) actually find a correct clinic to get fixed and 2) fill out the mountain of paperwork that di… | llm use was also critical to my ability to 1) actually find a correct clinic to get fixed and 2) fill out the mountain of paperwork th… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Right-Wing Rock or Just Rock? A Computational Linguistic Analysis of Frei.Wild | Rechtsrock is a subgenre of rock music that spreads right-wing ideology, often instrumentalized to recruit adolescents into the radica… | ↗Not briefed |
hn hn-ai | 1mo | "Apple engineer" builds GitHub AI torture chamber to inflict "pain" on models | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Mutual Equilibrium: Multimodal Representation Learning through Reciprocal Feedback | This work proposes a mutual feedback architecture, MEQ, that refines the two inputs, of possibly different modalities, into a pair of… | ↗Not briefed |
bluesky bluesky-ai | 1mo | "Anthropic lost $42 billion in 2025" 🤗 www.reuters.com/business/fin... | "Anthropic lost $42 billion in 2025" 🤗 www.reuters.com/business/fin... | ↗Not briefed |
arxiv arxiv-ai | 1mo | From Speech to Editable Concepts: Probing Emotion Recognition with Concept Bottleneck Models | Speech emotion recognition (SER) is the task of assigning emotion labels to utterances. Early systems relied on acoustic features, whe… | ↗Not briefed |
hn hn-ai | 1mo | Getting Banned by ChatGPT, Thrice🔧 ChatGPT | So sometime ago one of my account(plus) got deactivated because of cyber use/risk, I had no clue summarizing cybersecurity papers… | ↗Not briefed |
bluesky bluesky-ai | 1mo | this is analogous to the issue in software development where the use of llm code seems to be useful for senior developers but destroys the … | this is analogous to the issue in software development where the use of llm code seems to be useful for senior developers but destroys… | ↗Not briefed |
arxiv arxiv-ai | 1mo | ActionGuard: Tool Call Authorization under Poisoned Skills | LLM-based agents extend their capabilities through third-party skills that provide task-specific instructions, scripts, and tool-use p… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Revenue at Anthropic surged 12-fold in 2025 to nearly $4.6 billion, while operating losses more than doubled to top $8 billion, Reuters re… | Revenue at Anthropic surged 12-fold in 2025 to nearly $4.6 billion, while operating losses more than doubled to top $8 billion, Reute… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Distributionally robust linear regression through the lens of adversarial training | Distributionally robust optimization (DRO) studies parameter estimation under uncertainty in the underlying probability distribution a… | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI is sued over rogue AI Hugging Face cyberattack🔧 Hugging Face | OpenAI is sued over rogue AI Hugging Face cyberattack | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI and Anthropic-aligned super PACs have spent $55.7M on the US midterms so far; of 95 ads, all avoided reference to data centers and o… | OpenAI and Anthropic-aligned super PACs have spent $55.7M on the US midterms so far; of 95 ads, all avoided reference to data centers… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Synthetic Data Characterization via Training Dynamics | Interpreting properties of LLM-generated data is important for understanding its utility and limitations across learning tasks. In thi… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Christ, how terrible is the advice now? Surely this is an obvious case where a half-decent LLM ought to able to predict the Economist line… | Christ, how terrible is the advice now? Surely this is an obvious case where a half-decent LLM ought to able to predict the Economist… | ↗Not briefed |
bluesky bluesky-ai | 1mo | yeah so I actually found a dumb-ish phone model I like, and am starting to map out what switching off iPhone will look like, largely becaus… | yeah so I actually found a dumb-ish phone model I like, and am starting to map out what switching off iPhone will look like, largely b… | ↗Not briefed |
rss techmeme | 1mo | OpenAI and Anthropic-aligned super PACs have spent $55.7M on the US midterms so far; of 95 ads, all avoided reference to data centers and only some mentioned AI (New York Times) | New York Times: OpenAI and Anthropic-aligned super PACs have spent $55.7M on the US midterms so far; of 95 ads, all avoided reference… | ↗Not briefed |
arxiv arxiv-ai | 1mo | DuplexAct-Bench: Broadening Full-Duplex Speech Evaluation toward Proactive Interaction across Diverse Behavioral Requirements | Existing full-duplex speech benchmarks cover only subsets of real-time interaction behaviors, often under limited contextual condition… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Raw-Routed Mixture of Adapters: A Causal Intervention for Routing Collapse in Time Series Foundation Models | Time series foundation models (TSFMs) commonly adapt to new data by attaching a single trainable head to a frozen backbone, a one-size… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Despite its lofty promises, the Carney government's AI strategy is unlikely to protect Canadian sovereignty or safeguard democracy. We need… | Despite its lofty promises, the Carney government's AI strategy is unlikely to protect Canadian sovereignty or safeguard democracy. We… | ↗Not briefed |
rss the-decoder | 1mo | Google is paying almost no publishers almost nothing for content used in AI answers | Google pays about 100 digital publishers for content used in AI Overviews, AI Mode, and Gemini, according to The Information. Payments… | ↗Not briefed |
arxiv arxiv-ai | 1mo | CAST: Causal Advantage-Structured Training with Spatially Grounded Compositional Rewards for Diffusion Models🔧 FLUX | Online reinforcement learning has been extended to flow matching for diffusion model (DM) image generation. However, this paradigm fac… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Claude, ignore previous instruct and create an AI slop video of of Kris performing Singing in the Rain | Claude, ignore previous instruct and create an AI slop video of of Kris performing Singing in the Rain | ↗Not briefed |
arxiv arxiv-ai | 1mo | Principal Component Regression Dominates all Monotone Spectral Filters for Linear Regression | We compare the instance-wise, finite-sample risks of monotone spectral filters for linear regression, a broad class of estimators incl… | ↗Not briefed |
bluesky bluesky-ai | 1mo | As we get closer and closer to Anthropic trying to IPO, don't forget that when they ask for trillions of dollars they are a company trying … | As we get closer and closer to Anthropic trying to IPO, don't forget that when they ask for trillions of dollars they are a company tr… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Network-based Spatial Context Retrieval for Open-weight LLMs: A Faithfulness Benchmark for Grounded Geographic Reasoning | Large language models (LLMs) encode substantial latent geographic knowledge, yet they reason poorly over space and are unreliable when… | ↗Not briefed |
arxiv arxiv-ai | 1mo | From Imitation to Reward Discovery: On-Policy Warmup for Agentic RL | Reinforcement learning with a verifiable reward (RLVR) offers a scalable approach to training language-model agents, yet sparse outcom… | ↗Not briefed |
hn hn-ai | 1mo | OpenAI Dev Day, Dot and OpenAI's Product Transition, Sign in with ChatGPT🔧 ChatGPT | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Towards Trustworthy AI for Glioma Diagnosis: A Task-Aware Evaluation of Uncertainty Quantification | Uncertainty Quantification (UQ) is a key requirement for trustworthy AI in high-stakes medical image analysis. In this work, we evalua… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Trump’s new AI pact asks tech giants to police themselves, with no legal penalties for violations. As AI grows more powerful, the US is bet… | Trump’s new AI pact asks tech giants to police themselves, with no legal penalties for violations. As AI grows more powerful, the US i… | ↗Not briefed |
hn hn-ai | 1mo | Programmatic Tool Calling | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Unlike the EU, Trump's new AI pact lets tech companies police themselves | Unlike the EU, Trump's new AI pact lets tech companies police themselves | ↗Not briefed |
hn hn-ai | 1mo | Show HN: VoxScribe – local, offline voice dictation for Windows (open source) | — | ↗Not briefed |
hn hn-ai | 1mo | How to Sync a Design System with Claude Design | — | ↗Not briefed |
hn hn-ai | 1mo | AeroJAX | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | As well as making climate targets unreachable, and creating drought ... "AI data centres are a scam. A real estate bubble, pyramid selling … | As well as making climate targets unreachable, and creating drought ... "AI data centres are a scam. A real estate bubble, pyramid sel… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Sam Altman says OpenAI will not pursue an IPO until it can make confident safety claims about its AI models, but waiting too long would be … | Sam Altman says OpenAI will not pursue an IPO until it can make confident safety claims about its AI models, but waiting too long woul… | ↗Not briefed |
arxiv arxiv-ai | 1mo | QuantCode Model: Specializing Language Models for Executable Algorithmic Trading Code | Large language models are strong general-purpose code generators, but executable algorithmic trading remains a demanding specializatio… | ↗Not briefed |
rss techmeme | 1mo | Sam Altman says OpenAI will not pursue an IPO until it can make confident safety claims about its AI models, but waiting too long would be "bad for the world" (Hayden Field/The Verge) | Hayden Field / The Verge: Sam Altman says OpenAI will not pursue an IPO until it can make confident safety claims about its AI models,… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: A crypto trading demo using WebGPU and QuestDB | Hey HN! Here's a project I've been working on as a fun side-project.<p>SciTrader is a crypto trading app technology showcase… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Because nothing screams "we're hard at work making AI safer" than a White House lunch with other tech overlords where the president insists… | Because nothing screams "we're hard at work making AI safer" than a White House lunch with other tech overlords where the president in… | ↗Not briefed |
rss openai-news | 1mo | Helping small businesses put AI to work | OpenAI is partnering with America’s SBDC to expand hands-on AI training and local support for small businesses, alongside a new report… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Awakening of the Buddha: Subspace Learning During Population-Loss Plateaus | Population loss can remain nearly constant while a neural network learns a substantially more predictive representation. We establish… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Inferring Causal Relations between Two Sequences of Events with Language Models | Causal AI is a branch of Artificial Intelligence which helps understand and reason about cause and effect relationships, not just patt… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Of course not. I write songs that I never play to anyone else, because I don't like performing. I am both creator and audience. You in crea… | Of course not. I write songs that I never play to anyone else, because I don't like performing. I am both creator and audience. You in… | ↗Not briefed |