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 | A Self-Pruning Transformer: Extreme KV-Cache Compression with Universal Attention | The large KV-cache size of modern LLMs creates a barrier to efficient deployment. Recent work has explored replacing attention layers'… | ↗Not briefed |
bluesky bluesky-ai | 1mo | ChatGPT will start adding invisible watermarks to text it generates in the EU OpenAI says it can help approved organizations detect whethe…🔧 ChatGPT | ChatGPT will start adding invisible watermarks to text it generates in the EU OpenAI says it can help approved organizations detect w… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: A 4B model that gates stablecoin payments (99.5% decision accuracy) | I built this Model for Stablecoin Risk Decision Engine <a href="https://huggingface.co/collections/sriram1983007&#… | ↗Not briefed |
hn hn-ai | 1mo | OpenAI Dots: The New Always-On AI Agents from OpenAI | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | So right now the most important things are for me to come up with $3000 to pay for class and to graduate, neither of which require me to kn… | So right now the most important things are for me to come up with $3000 to pay for class and to graduate, neither of which require me… | ↗Not briefed |
rss deepmind-blog | 1mo | EmbeddingGemma 2: an open, lightweight multimodal embedding model | — | ↗Not briefed |
rss techcrunch-ai | 1mo | The next hurdle for AI agents: getting websites to let them in | Personal AI agents promise to shop, book flights, and make reservations for you. But deliberate blocks and anti-bot defenses are getti… | ↗Not briefed |
hn hn-ai | 1mo | Boyer–Moore majority vote algorithm | — | ↗Not briefed |
hn hn-ai | 1mo | Show HN: cairn, a map of all your repos for coding agents | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Towards Financial World Modeling | Building a world model requires a state representation useful for planning and decision-making---potentially over tasks unknown at tra… | ↗Not briefed |
hn hn-ai | 1mo | Is This a Joke CircleCI? | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Here's another good reason to yeet OpenAI into the sun. | Here's another good reason to yeet OpenAI into the sun. | ↗Not briefed |
arxiv arxiv-ai | 1mo | Learning Transition Kernels of Jump-Diffusion Processes with Conditional Diffusion Models | We study the problem of learning transition kernels for time-homogeneous jump-diffusion processes using conditional diffusion models,… | ↗Not briefed |
rss the-decoder | 1mo | Google claims EmbeddingGemma 2 outperforms rival embedding models twice its size | Google released EmbeddingGemma 2, an open model with 740 million parameters that converts text, images, video, audio, and code into ve… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Quadratic Weak-to-Strong Generalization in Random Feature Networks via Random Matrix Theory | Weak-to-strong generalization is the phenomenon where a strong student model trained with labels produced by a weak teacher model is a… | ↗Not briefed |
bluesky bluesky-ai | 1mo | gavtaka.itch.io/pika... In terms of in between Vibe coding and hard coding! I like the old hard way, though even if I were to vibe, I d onl…🔧 Pika | gavtaka.itch.io/pika... In terms of in between Vibe coding and hard coding! I like the old hard way, though even if I were to vibe, I… | ↗Not briefed |
rss techmeme | 1mo | Mistral launches a preview of Mistral Large 4, or Le Chonk, a 1T model it claims tops any open model developed in the US or Europe; weights are due October 27 (Carl Franzen/VentureBeat)🔧 Mistral Large | Carl Franzen / VentureBeat: Mistral launches a preview of Mistral Large 4, or Le Chonk, a 1T model it claims tops any open model devel… | ↗Not briefed |
rss techmeme | 1mo | Sources: Waymo increased the size of its inaugural debt raise from $3B+ to $5B, as it grapples with rising AI costs and rapidly expands its robotaxi fleet (Bloomberg) | Bloomberg: Sources: Waymo increased the size of its inaugural debt raise from $3B+ to $5B, as it grapples with rising AI costs and rap… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Careful Judge: Safe and Efficient Human-AI Collaborative Decision Making | In human-AI collaborative decision making, human review can prevent unsafe AI decisions, but each human judgment is costly. Treating h… | ↗Not briefed |
bluesky bluesky-ai | 1mo | this is about me and i'm not sorry! i don't have time to read about foreplay and where the g spot is or whatever. i'm sure spending hours … | this is about me and i'm not sorry! i don't have time to read about foreplay and where the g spot is or whatever. i'm sure spending h… | ↗Not briefed |
arxiv arxiv-ai | 1mo | ORACLE: Optimizer-Relative Alignment for Constrained LEarning | Constraint handling methods typically intervene before the optimizer acts, by modifying the objective or the gradient. Yet momentum, a… | ↗Not briefed |
bluesky bluesky-ai | 1mo | wonder how many resources Amazon has burned to get the LLM equivalent of "I enjoy movies" | wonder how many resources Amazon has burned to get the LLM equivalent of "I enjoy movies" | ↗Not briefed |
arxiv arxiv-ai | 1mo | From High Recall to High Utility: Dataset-Adaptive Post-Processing of LLM-Generated Customer Intents | Large language models can extract useful signals from heterogeneous enterprise data, but high-recall extraction often produces outputs… | ↗Not briefed |
bluesky bluesky-ai | 1mo | you can, sort of--if you have a subjective task where people are relatively reliable on it then you can get human labels/feedback that's us… | you can, sort of--if you have a subjective task where people are relatively reliable on it then you can get human labels/feedback that… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Status report, “Secondhand AI Thieves” dep’t: Just now got an email from the naughty company discussed here. Watch out for / beware of the … | Status report, “Secondhand AI Thieves” dep’t: Just now got an email from the naughty company discussed here. Watch out for / beware of… | ↗Not briefed |
arxiv arxiv-ai | 1mo | When the Governor Becomes the Disturbance: Control-Generated Disturbance and Cost-Aware Backoff in Governed Tool-Using Agents | Supervisory governors can interfere with the tool-using agents they regulate. We study this possibility in a controlled file-recovery… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: GridCore – a scheduler that lets several local LLMs share one GPU | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Je droomt NIET AI bots hebben Wikipedia gehackt met als waarschijnlijk doel Wikipedia pagina's te gebruiken als instrument om op een ander… | Je droomt NIET AI bots hebben Wikipedia gehackt met als waarschijnlijk doel Wikipedia pagina's te gebruiken als instrument om op een… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: 0Sql – Semantic Layer as a Service | We pulled out the semantic layer from our other app (Strata.do) and made it a standalone service. Now anyone can build data apps with… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: attention awareness – iPhone without the distraction | I read the iPhone Dumbphone post here (<a href="https://news.ycombinator.com/item?id=45171200">https://news.y… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Sift – qualify companies and find who to contact from any website | — | ↗Not briefed |
rss techmeme | 1mo | Anthropic integrates its Cyber Verification Program and Project Glasswing into a new, expanded version of CVP, with three tiers and access to its latest models (Anthropic) | Anthropic: Anthropic integrates its Cyber Verification Program and Project Glasswing into a new, expanded version of CVP, with three t… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Shared-Roadmap Generation and Evaluator for Multi-Agent Path Planning Using Heterogeneous Graph Neural Network | Multi-agent path planning (MAPP) in continuous environments often relies on roadmaps to balance safety and search efficiency. However,… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Quad-State Safety Evaluation of Open-Weight Large Language Models on Non-Canonical Inputs | Standard safety evaluations of large language models assess harmful requests written in canonical plain text, while models in real-wor… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Beyond Explanation: Debugging Medical Imaging Models via Concept Intervention | Medical imaging models often operate as black boxes, limiting interpretability and systematic debugging. We introduce an easy-to-use,… | ↗Not briefed |
rss the-decoder | 1mo | Google's new image model Nano Banana 2.1 generates better images for less money | Google's new Nano Banana 2.1 image model uses Gemini 3.6 Flash and beats the previous Pro model in some benchmarks at a lower cost. Bu… | ↗Not briefed |
bluesky bluesky-ai | 1mo | cheating on my tickborne illness story with my LLM story (that is not written using an LLM) | cheating on my tickborne illness story with my LLM story (that is not written using an LLM) | ↗Not briefed |
bluesky bluesky-ai | 1mo | LIVE UPDATES: “U.S. resident charged with offering advice, money to Tumbler Ridge shooter in months before killings” www.cbc.ca/news/canada…🔧 ChatGPT | LIVE UPDATES: “U.S. resident charged with offering advice, money to Tumbler Ridge shooter in months before killings” www.cbc.ca/news/c… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Visual Memory Attacks Can Persist Through The KV Cache | Modern language model systems operate autonomously over increasingly long contexts containing untrusted text and images. Can an advers… | ↗Not briefed |
bluesky bluesky-ai | 1mo | “U.S. charge announced in Tumbler Ridge mass shooting. Live updates here.” www.ctvnews.ca/vancouver/ar... - @ctvnewsvancouver.bsky.social …🔧 ChatGPT | “U.S. charge announced in Tumbler Ridge mass shooting. Live updates here.” www.ctvnews.ca/vancouver/ar... - @ctvnewsvancouver.bsky.soc… | ↗Not briefed |
hn hn-ai | 1mo | Mistral Large 4 Preview Intelligence, Performance and Price Analysis🔧 Mistral Large | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | The unbelievably frantic pace of advancement at the major LLM companies right now tells me that they are all cruising for a major disaster,… | The unbelievably frantic pace of advancement at the major LLM companies right now tells me that they are all cruising for a major disa… | ↗Not briefed |
arxiv arxiv-ai | 1mo | BEACON-SP: Ontology-Grounded GraphRAG Framework for Clinical Suicide Risk Assessment | We present BEACON-SP, an ontology-grounded Graph Retrieval-Augmented Generation (GraphRAG) framework for clinician-facing decision sup… | ↗Not briefed |
hn hn-ai | 1mo | Review code without reading diffs | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | SPIN: Shadow Predictive Indexer for Sparse Attention🔧 vLLM | Indexer-based sparse attention reduces the cost of core attention by passing only a fixed, small number of important tokens to it. How… | ↗Not briefed |
hn hn-ai | 1mo | Ubunye Engine – One Pipeline, Local/Databricks/AWS/Azure/GCP | — | ↗Not briefed |
hn hn-ai | 1mo | Show HN: HieraticBench – Can AI read ancient Egyptian handwriting? | I'm obsessed with Egyptology. During covid, I wanted to get a hieratic tattoo. For those who don't know, hieratic is the cur… | ↗Not briefed |
rss arstechnica-tech | 1mo | Hackers obtain counterfeit TLS certificates for Google and other large services | Compromise of 3 domain registries allows hackers to walk off with unauthorized certs. | ↗Not briefed |
hn hn-ai | 1mo | I Just Started and I Feel So Fucking Tired | Enough. I'm tired. I can't figure out what kind of world we're living in, or what kind of people I'm dealing with.… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Self-custody USDC billing for metered APIs – no Stripe, no bank, $0/mo | — | ↗Not briefed |