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 | A Tale of Two Programmers | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Executing Causal Structure Learning with Linear-Attention Transformers | Transformers can execute algorithms on data given in their input. We ask whether they can do the same for causal discovery. We study a… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Kernel Autoresearch for Open-Ended Model Discovery | Kernels encode the inductive bias of a wide range of machine learning models, yet automated kernel design faces a fundamental dilemma.… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Safe Meta-Policy Design with Risk Control | Models can be retrained as new data arrive, but deploying every new version risks replacing a good policy with a worse one. We study h… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Someone used a research paper to set up a "Saw"-inspired AI "torture" chamber. Then, users on X started the dumbest debate about "model wel… | Someone used a research paper to set up a "Saw"-inspired AI "torture" chamber. Then, users on X started the dumbest debate about "mode… | ↗Not briefed |
bluesky bluesky-ai | 1mo | ChatGPT for Teens: "unacceptable risk" Common Sense Media calls OpenAI’s ChatGPT for Teens an “unacceptable risk” due to unreliable parent…🔧 ChatGPT | ChatGPT for Teens: "unacceptable risk" Common Sense Media calls OpenAI’s ChatGPT for Teens an “unacceptable risk” due to unreliable p… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Featured on the Home Page of Model Society - the pre-eminent source for fine art nudes. Thank you Model Society - never porn, never sexual… | Featured on the Home Page of Model Society - the pre-eminent source for fine art nudes. Thank you Model Society - never porn, never s… | ↗Not briefed |
bluesky bluesky-ai | 1mo | it's funny bc i hate 99% of LLM hype sooooo much but when i see stuff advocating for it to just be disappeared i'm like. well, that's not g… | it's funny bc i hate 99% of LLM hype sooooo much but when i see stuff advocating for it to just be disappeared i'm like. well, that's… | ↗Not briefed |
arxiv arxiv-ai | 1mo | RoboQuest: Generalist Physical Agents that Search, Inspect and Test | Recent advances in multimodal foundation models have made them capable generalist physical agents for a range of manipulation tasks. H… | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI says teens use ChatGPT for under 15 minutes a day as worries over risks grow reut.rs/4ja6H1I🔧 ChatGPT | OpenAI says teens use ChatGPT for under 15 minutes a day as worries over risks grow reut.rs/4ja6H1I | ↗Not briefed |
arxiv arxiv-ai | 1mo | OrBIT: Structure-Guided Embedding Compression | Embedding tables are among the largest components of modern language models. Most compression methods fix a coding geometry such as co… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Anthroptic and OpenAI repteatedly complain about China "stealing" their models. Yes, I believe so too. This is happening. But fuck you Ope… | Anthroptic and OpenAI repteatedly complain about China "stealing" their models. Yes, I believe so too. This is happening. But fuck yo… | ↗Not briefed |
bluesky bluesky-ai | 1mo | ngl having a macbook i realize just how much of what ppl said was bullshit its an eye opening experience. esp since the llm age where i can… | ngl having a macbook i realize just how much of what ppl said was bullshit its an eye opening experience. esp since the llm age where… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Boosting and the Expressive Power of Simple Weak Learners via the $γ$-VC Dimension | Boosting converts weak hypotheses with a small edge over random guessing into highly accurate predictors, but the expressive power of… | ↗Not briefed |
hn hn-ai | 1mo | LLM-OpenAI-Decisions 0.1a0 | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | ResidualQuant: KV Cache Quantization for Looped Transformers with 2-Bit Residuals | Looped Transformers improve parameter efficiency by repeatedly applying shared Transformer blocks over multiple recurrent loops, incre… | ↗Not briefed |
hn hn-ai | 1mo | OpenAI says teens use ChatGPT for under 15 minutes a day🔧 ChatGPT | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Continual Learning without Continual Training | Continual learning requires models to adapt to new domains and new classes while retaining prior knowledge. Many existing methods rely… | ↗Not briefed |
bluesky bluesky-ai | 1mo | basically im ngl if openai had decided to partner with universities to allow phd students access to their frontier models & tooling for fre… | basically im ngl if openai had decided to partner with universities to allow phd students access to their frontier models & tooling fo… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I sincerely doubt that open weight Mythos in April would have been fine. But GLM 5.3 has clearly been fine so far and Anthropic has been ve… | I sincerely doubt that open weight Mythos in April would have been fine. But GLM 5.3 has clearly been fine so far and Anthropic has be… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Document-Level Text Simplification in Estonian Using Large Language Models | Document-level text simplification involves transformations that go beyond sentence-internal edits, addressing discourse coherence, an… | ↗Not briefed |
arxiv arxiv-ai | 1mo | TaoD2C-Bench: Benchmarking MLLMs for Industrial UI Code Generation Beyond Visual Fidelity🔧 Modal | A key challenge for multimodal large language models (MLLMs) is moving beyond visual recognition to constraint-aware cross-modal reaso… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Aqui traduzido e em Português, pra que a mensagem fique bem clara. tecnoblog.net/noticias/mun... | Aqui traduzido e em Português, pra que a mensagem fique bem clara. tecnoblog.net/noticias/mun... | ↗Not briefed |
bluesky bluesky-ai | 1mo | (Yes I know that modern search engines already incorporated things that are meaningfully called LLMs, what I really mean is "LLM tuned for … | (Yes I know that modern search engines already incorporated things that are meaningfully called LLMs, what I really mean is "LLM tuned… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Que ninguém diga depois que não sabia, que os riscos não estavam claros, que ele não era tão transparente assim. É impossível ser mais tr… | Que ninguém diga depois que não sabia, que os riscos não estavam claros, que ele não era tão transparente assim. É impossível ser ma… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Input-Blind Controls Produce Substantial Oracle Headroom for Layer Programs in Multiple-Choice Evaluation | Adaptive computation aims to improve language-model inference by tailoring execution to each input. For layer programs, oracle evaluat… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Temporally Interpretable Differentiable Decision Trees | Interpretability offers a solution to safe autonomy by providing transparency into an agent's underlying decision-making model. Within… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Koopman Observers for Diffusion Acceleration: Correcting Feature Forecasts with Shallow Measurements | Feature caching accelerates diffusion sampling by replacing expensive network evaluations with predictions from previously computed ac… | ↗Not briefed |
bluesky bluesky-ai | 1mo | ChatGPT's teen safeguards failed to alert parents during suicide conversations, report finds🔧 ChatGPT | ChatGPT's teen safeguards failed to alert parents during suicide conversations, report finds | ↗Not briefed |
bluesky bluesky-ai | 1mo | Pleased to announce that I am launching AI Explained, a weekly column on how to understand AI. The first edition today is all about OpenAI'… | Pleased to announce that I am launching AI Explained, a weekly column on how to understand AI. The first edition today is all about Op… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Anthropic will never ever admit to reality in this case because they are all in a cult that tells them they cant admit to this reality. | Anthropic will never ever admit to reality in this case because they are all in a cult that tells them they cant admit to this reality. | ↗Not briefed |
bluesky bluesky-ai | 1mo | Anthropic will never ever admit to reality in this case because they are all in a cult that tells them they cant admit to this reality. | Anthropic will never ever admit to reality in this case because they are all in a cult that tells them they cant admit to this reality. | ↗Not briefed |
arxiv arxiv-ai | 1mo | Pathwise Information Certificates for Decentralized Adaptive Sensing | We study decentralized adaptive sensing, where multiple agents choose measurements from evolving local beliefs while exchanging inform… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Tengo que hacer un trabajo con un LLM y he elegido a Alan Turing porque (1) siempre está en mi corazón y (2) es muy gracioso. | Tengo que hacer un trabajo con un LLM y he elegido a Alan Turing porque (1) siempre está en mi corazón y (2) es muy gracioso. | ↗Not briefed |
bluesky bluesky-ai | 1mo | I think this is right, and I will add to it that part of why it's galling is that it's clear OpenAI doesn't give a shit about math or the m… | I think this is right, and I will add to it that part of why it's galling is that it's clear OpenAI doesn't give a shit about math or… | ↗Not briefed |
arxiv arxiv-ai | 1mo | ORDERS: An Empirical Study of Norm-Rank Aggregation for Personalized Federated Learning | Personalized federated learning combines shared representations with client-specific predictors, but the contribution of a server weig… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Open-MMUnlearning: Unifying Methods and Evaluation for MLLM Unlearning | As multimodal large language models (MLLMs) become more capable and widely deployed, concerns about privacy and safety have become inc… | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI might say that they’re done spending hundreds of millions on elections to block regulation - but we shouldn’t take them at their wor… | OpenAI might say that they’re done spending hundreds of millions on elections to block regulation - but we shouldn’t take them at thei… | ↗Not briefed |
rss techmeme | 1mo | OpenAI's first report on teen ChatGPT use says teens spend under 15 minutes a day on average on the chatbot, with less than 2% using it for 3+ hours straight (Reuters)🔧 ChatGPT | Reuters: OpenAI's first report on teen ChatGPT use says teens spend under 15 minutes a day on average on the chatbot, with less than 2… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Or I could have just searched Google Maps the old-fashioned way and ticked a filter box for "outdoor dining" because actually modern search… | Or I could have just searched Google Maps the old-fashioned way and ticked a filter box for "outdoor dining" because actually modern s… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Si une voiture court un 100m elle ne va pas remporter le titre mondial même si elle arrive en premier. | Si une voiture court un 100m elle ne va pas remporter le titre mondial même si elle arrive en premier. | ↗Not briefed |
arxiv arxiv-ai | 1mo | MIRA: A Musical Intent Refinement Agent for Aligning Text-to-Music Generation with User Intent🔧 Suno | Text-to-music systems produce increasingly convincing audio, yet evaluation reveals little about whether the result matches user inten… | ↗Not briefed |
hn hn-ai | 1mo | New Parking Meter Deal Gives Chicago Share of Profits for First Time Since 2008 | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | For when it's removed | For when it's removed | ↗Not briefed |
hn hn-ai | 1mo | Show HN: I made a desktop pet that uses LLM for voice and Jev for poses | This desktop pet uses Gemini Live and GPT realtime for realtime conversation, and jev for model pose control. Check the github for how… | ↗Not briefed |
arxiv arxiv-ai | 1mo | The Handover Problem: Governing Autonomy Transitions in Human-AI Collaboration | Human-machine systems rarely operate at a fixed level of AI autonomy. As operators and AI systems collaborate over time, control must… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Measurement-Efficient Differentiable Quantum Architecture Search for Combinatorial Optimization | Differentiable quantum architecture search (DQAS) is a promising framework for the automated design of quantum circuits, particularly… | ↗Not briefed |
arxiv arxiv-ai | 1mo | AutoAdapt: Automatic Domain Discovery Enables Low-Cost Extensibility | Instruction-tuned models are deployed into environments where domains are heterogeneous and evolve, yet adding new domains or data typ… | ↗Not briefed |
hn hn-ai | 1mo | NullMoth Nvidia Driver with Metal for macOS 15 | — | ↗Not briefed |
hn hn-ai | 1mo | Show HN: See when TV shows are ready to binge | Made this for impatient people like me, who have trouble dealing with the torturous drip feed of one new TV series episode per week, a… | ↗Not briefed |