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 | 1mo | Même s'il y avait du plagiat. Même s'il y avait de l'utilisation de LLM. RIEN ne justifie un harcèlement. RIEN. | Même s'il y avait du plagiat. Même s'il y avait de l'utilisation de LLM. RIEN ne justifie un harcèlement. RIEN. | ↗Not briefed |
bluesky bluesky-ai | 1mo | ‘By Jensen’s logic we ought at this point be shutting down OpenAI. Their software has repeatedly been shown to be reckless, and because the… | ‘By Jensen’s logic we ought at this point be shutting down OpenAI. Their software has repeatedly been shown to be reckless, and becaus… | ↗Not briefed |
bluesky bluesky-ai | 1mo | that imo is even more of a red flag of LLM usage than em dashes because em dashes were already used in writing long before LLMs became a th… | that imo is even more of a red flag of LLM usage than em dashes because em dashes were already used in writing long before LLMs became… | ↗Not briefed |
bluesky bluesky-ai | 1mo | What drives me crazy is that the actual issue is automating anomaly detection over a massive amount of genome data - a specific problem LLM… | What drives me crazy is that the actual issue is automating anomaly detection over a massive amount of genome data - a specific proble… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Let Training Guide Selection: Online Synthetic Data Filtering via Real-Anchored Utility | Synthetic data can scale training supervision when real-world data are limited, but noise and distribution mismatch can reduce its val… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Armada – Every Claude Code and Codex session and limit, on one screen | — | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Enclawed – A hard-fork hardening framework for AI agent gateways | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | « Après le débat sur les risques des IA, un autre sur leur “conscience” se profile dans la Silicon Valley » | « Après le débat sur les risques des IA, un autre sur leur “conscience” se profile dans la Silicon Valley » | ↗Not briefed |
arxiv arxiv-ai | 1mo | From Interests to Semantic IDs: Retrieval-Grounded Credit Assignment for Generative Recommendation | Semantic IDs (SIDs) encode each catalog item as a short token sequence, enabling generative recommenders to predict the next item auto… | ↗Not briefed |
bluesky bluesky-ai | 1mo | IT業界は、すでに、プログラムは自分の手で書くよりLLMに書かせる方がいけてる、って雰囲気はあるね。なんでかと言うと、自分の手でコード書いたり外注さんとか部下に書いてもらうより、LLMに書かせた方が精度それなりで圧倒的に早くて安いから。手で書くのはコストがかかって遅いのに、なん… | IT業界は、すでに、プログラムは自分の手で書くよりLLMに書かせる方がいけてる、って雰囲気はあるね。なんでかと言うと、自分の手でコード書いたり外注さんとか部下に書いてもらうより、LLMに書かせた方が精度それなりで圧倒的に早くて安いから。手で書くのはコストがかかって遅い… | ↗Not briefed |
bluesky bluesky-ai | 1mo | The link to this is going straight into my LLM policy for the course I'm currently developing! | The link to this is going straight into my LLM policy for the course I'm currently developing! | ↗Not briefed |
bluesky bluesky-ai | 1mo | Something like this: “To whom it may concern, Our rogue AI has now hacked your secure government accounts. Sorry for the inconvenience. �… | Something like this: “To whom it may concern, Our rogue AI has now hacked your secure government accounts. Sorry for the inconvenien… | ↗Not briefed |
bluesky bluesky-ai | 1mo | To cycle back to HuggingFace, one of the many factors I think OpenAI have been coy about is "how did you not notice you were burning throug… | To cycle back to HuggingFace, one of the many factors I think OpenAI have been coy about is "how did you not notice you were burning t… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Diverse Geometries, Frozen Weights: Robust Heterogeneous Treatment-Effect Estimation via Causal Expert Ensembles | Estimating heterogeneous treatment effects from observational data is difficult because the most appropriate inductive bias varies wit… | ↗Not briefed |
bluesky bluesky-ai | 1mo | The thing that I think a lot of people don’t get about LLM-based hacking systems is that they don’t need to be smart or sophisticated — jus… | The thing that I think a lot of people don’t get about LLM-based hacking systems is that they don’t need to be smart or sophisticated… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Learning Better Reasoning for Generative Recommendation with Semantic IDs | Generative recommendation reformulates item retrieval as sequence generation, allowing a unified model to directly generate the next i… | ↗Not briefed |
bluesky bluesky-ai | 1mo | AI panic ignores “the climate and ecological crisis, the rising fascism that is borne out of climate politics, the making of climate refuge… | AI panic ignores “the climate and ecological crisis, the rising fascism that is borne out of climate politics, the making of climate r… | ↗Not briefed |
hn hn-ai | 1mo | Paperclip: Open-source orchestration for teams of AI agents | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | the easiest, simplest, and best LLM policy is still, and will forever remain, "no, fuck you, don't use it". | the easiest, simplest, and best LLM policy is still, and will forever remain, "no, fuck you, don't use it". | ↗Not briefed |
bluesky bluesky-ai | 1mo | “They are claiming that they have the power to end humanity, but at the same time can’t get AI to do a lot of major things that they’re pus… | “They are claiming that they have the power to end humanity, but at the same time can’t get AI to do a lot of major things that they’r… | ↗Not briefed |
bluesky bluesky-ai | 1mo | One additional thing to add (I have some personal experience with this issue...) - Huang is obviously an extremely key person in AI, but he… | One additional thing to add (I have some personal experience with this issue...) - Huang is obviously an extremely key person in AI, b… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Week after week democracy now breathlessly push this doomer booster narrative the AI companies want the media to parrot. Incredibly disappo…🔧 ChatGPT | Week after week democracy now breathlessly push this doomer booster narrative the AI companies want the media to parrot. Incredibly di… | ↗Not briefed |
bluesky bluesky-ai | 1mo | checking in on Fluxer (I'm still not going to use it). They did a big "v2" change a while ago that changed the LLM policy. what strikes me… | checking in on Fluxer (I'm still not going to use it). They did a big "v2" change a while ago that changed the LLM policy. what strik… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Jev-phone – driving iOS and Android apps with jev | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | amid NYC Climate Week and the UN General Assembly (and months of climate catastrophe from Nepal to California), climate organizers from acr… | amid NYC Climate Week and the UN General Assembly (and months of climate catastrophe from Nepal to California), climate organizers fro… | ↗Not briefed |
arxiv arxiv-ai | 1mo | World Action Agent: Harnessing VLMs for Robot Manipulation via World Action Rehearsal | General-purpose vision-language models (VLMs) bring broad knowledge and spatial reasoning to robot manipulation, yet existing systems… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Anthropic says its biology lab has already found something big The AI giant announced that this lab has already made what it believes is a… | Anthropic says its biology lab has already found something big The AI giant announced that this lab has already made what it believes… | ↗Not briefed |
arxiv arxiv-ai | 1mo | ADATEX4D: adaptive texture capacity allocation for 4D gaussian splatting | Textured Gaussians improve local appearance capacity, but assigning the same texture resolution to every primitive wastes storage on l… | ↗Not briefed |
hn hn-ai | 1mo | Local JEV, fine tuned in < 30 mins on Mac Air | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | A Contraction Framework for Stochastic Operators with Bootstrapping: Application to TD Learning | Many iterative algorithms rely on bootstrapping. A variable is updated using a second, frozen copy as a target, which is periodically… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Beyond Average Safety: Chance-Constrained LLM Fine-tuning | Fine-tuning large language models on new objectives can improve helpfulness, instruction following, or domain-specific performance, bu… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Not All Confusion Is Equal: A Source-Aware Uncertainty Diagnosis for Fine-Grained Aircraft Detection | Fine-grained object detectors are commonly evaluated with confusion matrices, which show where the model is confused but not why, nor… | ↗Not briefed |
hn hn-ai | 1mo | Make Videos Programmatically with React | — | ↗Not briefed |
hn hn-ai | 1mo | What Australia's Hidden Caves Can Teach Us About Martian Terrain | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Multi-Dimensional Matching | We study a matching mechanism where agents and objects are described by features rather than complete rankings. A single spectral proj… | ↗Not briefed |
bluesky bluesky-ai | 1mo | AI agents at OpenAI, Anthropic, Meta and Google have gone rogue in recent months, hacking into other companies’ corporate systems and, in t… | AI agents at OpenAI, Anthropic, Meta and Google have gone rogue in recent months, hacking into other companies’ corporate systems and,… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Are My ETFs Overvalued?" | I built ETF Copilot (<a href="https://etf-copilot.com/go/show-hn" rel="nofollow">https://etf-copilot.com… | ↗Not briefed |
bluesky bluesky-ai | 1mo | In my upcoming Thought Experiment Revolution, I describe Large Language Models as "massive misinformation machines." My copyeditor has peti… | In my upcoming Thought Experiment Revolution, I describe Large Language Models as "massive misinformation machines." My copyeditor has… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Augur: A Synthetic Decision Lab for Rehearsing Reactions to Product and Policy Changes | Before a product or policy change ships, the question that matters is how people will react to it. Augur rehearses that reaction offli… | ↗Not briefed |
rss techmeme | 1mo | At a White House summit, Xi Jinping says the US and China share the "capability and responsibility" to manage AI for good and calls for healthy competition (Financial Times) | Financial Times: At a White House summit, Xi Jinping says the US and China share the “capability and responsibility” to manage AI for… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Tracking States or Tracking Cosets? An Algebraic Account of Learned State Tracking | State tracking requires composing a sequence of updates, but accuracy alone does not reveal what a model has learned. We study neural… | ↗Not briefed |
bluesky bluesky-ai | 1mo | deepseek and friends are just so insanely cheap nowadays, you can get a long way with smart orchestration | deepseek and friends are just so insanely cheap nowadays, you can get a long way with smart orchestration | ↗Not briefed |
bluesky bluesky-ai | 1mo | Yeah I might have taken too much of a reactionary attitude against AI safety and alignment talks when we often just trust the model too muc… | Yeah I might have taken too much of a reactionary attitude against AI safety and alignment talks when we often just trust the model to… | ↗Not briefed |
arxiv arxiv-ai | 1mo | ENDOPROMPT: Victim-Side Pseudo-References for Utility Degradation | Prompt injection can degrade benign task performance without eliciting harmful content. Yet many attack objectives depend on task labe… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Neuro-symbolic AI for Industrial Configuration | Large Language Models (LLMs) have shown impressive performance on a wide range of generative tasks. Yet their probabilistic nature mak… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Should be noted that explanations that appeal to what exactly motivated the agents etc are highly speculative and are due to sources close … | Should be noted that explanations that appeal to what exactly motivated the agents etc are highly speculative and are due to sources c… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Error- and Prediction-Driven Motor Learning in the Cortico-Cerebellar Loop | Robust control under delayed sensory feedback remains a key challenge in both robotics and neuroscience. Classical cerebellar models e… | ↗Not briefed |
arxiv arxiv-ai | 1mo | MF-SCBO : Multi-fidelity Scalable Constrained Bayesian Optimization | Many real-world optimization problems rely on expensive simulations or experiments, making the efficient use of available data essenti… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Mind What Matters for Reasoning: Aligning Cross-Modal Attention via Selective Probability Mass Concentration🔧 Modal | Multimodal large language models (MLLMs) achieve strong performance on visual reasoning tasks, yet remain prone to hallucinations and… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Why hasn't anyone CRIMINALLY CHARGED THEM WITH HACKING? Or Anthropic? | Why hasn't anyone CRIMINALLY CHARGED THEM WITH HACKING? Or Anthropic? | ↗Not briefed |