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 | Gap – Graph-as-Policy | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | @danielstone.bsky.social proposes making disclosure a legal duty, and that firms such as OpenAI must hand affected organizations the logs a… | @danielstone.bsky.social proposes making disclosure a legal duty, and that firms such as OpenAI must hand affected organizations the l… | ↗Not briefed |
bluesky bluesky-ai | 1mo | If an LLM causes a centaur-bot to chainsaw a tree in a forest but there's noone there to hear it, does it still make a noise as it falls? | If an LLM causes a centaur-bot to chainsaw a tree in a forest but there's noone there to hear it, does it still make a noise as it fal… | ↗Not briefed |
hn hn-ai | 1mo | Ask HN: Why would I use Openrouter over something like DeepInfra directly?🔧 OpenRouter🔧 Baseten | As far as I understand it, OpenRouter is just routing to providers such as DeepInfra, Baseten, Fireworks, Together, etc, who already p… | ↗Not briefed |
bluesky bluesky-ai | 1mo | "You can decide never to use an AI agent and still have to deal with the work it creates." AI efficiency claims, Stone argues, risk "mistak… | "You can decide never to use an AI agent and still have to deal with the work it creates." AI efficiency claims, Stone argues, risk "m… | ↗Not briefed |
arxiv arxiv-ai | 1mo | A generative-informed neuro-symbolic framework for syntactic ambiguity resolution: Evidence from Arabic DPs | Syntactic ambiguity poses a persistent challenge for Arabic NLP, particularly in morphologically rich nominal constructions where mult… | ↗Not briefed |
bluesky bluesky-ai | 1mo | SocArXiv now has a limit of two papers per week. If authors submit more than that we assume they are LLM-generated and reject them all. Nor… | SocArXiv now has a limit of two papers per week. If authors submit more than that we assume they are LLM-generated and reject them all… | ↗Not briefed |
rss bing-news-ai | 1mo | Palantir billionaire Peter Thiel has 42% of his portfolio in these 2 artificial intelligence stocks | The famed entrepreneur and investor targeted two companies pursuing different aspects of the AI market. | ↗Not briefed |
bluesky bluesky-ai | 1mo | its literally just chatgpt running customer service on the website, if the delivery guy doesnt take a picture of the package on ur doorstep…🔧 ChatGPT | its literally just chatgpt running customer service on the website, if the delivery guy doesnt take a picture of the package on ur doo… | ↗Not briefed |
bluesky bluesky-ai | 1mo | The warning the company eventually sent was to a general public inbox. Stone notes OpenAI has roughly 75 staff in Australia and a lobbying … | The warning the company eventually sent was to a general public inbox. Stone notes OpenAI has roughly 75 staff in Australia and a lobb… | ↗Not briefed |
hn hn-ai | 1mo | Matryoshka: MCP for token-efficient large document analysis via REPL | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Autoregressive Differentiable Method for Integer Programming | We introduce an autoregressive differentiable method to solve 0-1 integer programs. We fix an arbitrary order of the binary variables… | ↗Not briefed |
bluesky bluesky-ai | 1mo | The timeline: The agent got in June 18. OpenAI says it noticed Aug. 11, 54 days later. It told Australia Sept. 10. That's 84 days, "almost … | The timeline: The agent got in June 18. OpenAI says it noticed Aug. 11, 54 days later. It told Australia Sept. 10. That's 84 days, "al… | ↗Not briefed |
arxiv arxiv-ai | 1mo | CriticHack: Evaluating Visual Rewards Under Robot Policy Optimization | Learned visual reward models are increasingly used to optimize robot policies, yet a reward model can score an execution that acts on… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I would say that meaning is something like "the predicted effect that text will have on the recipient." In the case of an LLM writing code,… | I would say that meaning is something like "the predicted effect that text will have on the recipient." In the case of an LLM writing… | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI's agent went looking for Australian medicine spending figures and got write access to a Medicare statistics server. Daniel Stone of … | OpenAI's agent went looking for Australian medicine spending figures and got write access to a Medicare statistics server. Daniel Ston… | ↗Not briefed |
rss techmeme | 1mo | Source: some of the information that the three OpenAI employees allegedly mishandled pertained to OpenAI's infrastructure architecture (Rachel Metz/Bloomberg) | Rachel Metz / Bloomberg: Source: some of the information that the three OpenAI employees allegedly mishandled pertained to OpenAI's in… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Learning What to Investigate Next: Meta-Reasoning for Long-Horizon Research Agents | Long-horizon research agents must decide both how to investigate and what to investigate next as evidence accumulates. This is hard to… | ↗Not briefed |
hn hn-ai | 1mo | Gboard Conveyor Belt Version | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | BaCP: Backbone Contrastive Pruning for Preserving Representations in Extremely Sparse Neural Networks | Unstructured pruning at extreme sparsity often suffers from representational collapse, causing sharp drops in accuracy. To address thi… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Hypothesis-guided discovery of cognitive algorithms via program refinement | Developing cognitive models of algorithmic reasoning from behavioral data is a central problem in cognitive science that challenges cu… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Worse, a wannabe script-kiddy no longer needs any actual knowledge of cybersec to hack - they just need a GPU, and a jailbroken LLM install… | Worse, a wannabe script-kiddy no longer needs any actual knowledge of cybersec to hack - they just need a GPU, and a jailbroken LLM in… | ↗Not briefed |
hn hn-ai | 1mo | H-HPU: Hexagonal NoC and Closed-Loop Control for Relational AI | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | An LLM doesn't need to be smarter than or even as smart as a human. It just has to try every conceivable option for penetrating a system u… | An LLM doesn't need to be smarter than or even as smart as a human. It just has to try every conceivable option for penetrating a sys… | ↗Not briefed |
bluesky bluesky-ai | 1mo | this is it. the difference between you and me and an llm is that an llm produces language in response to a prompt, and you and I *use langu… | this is it. the difference between you and me and an llm is that an llm produces language in response to a prompt, and you and I *use… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Instance-Dependent Regret for CMDPs with Step-Wise Constraints | We study online learning in episodic tabular constrained Markov decision processes with step-wise safety constraints. In such a settin… | ↗Not briefed |
bluesky bluesky-ai | 1mo | llms are not neutral arbiters of information and any llm corporation that's aware the us president is prompting it for answers can indirect… | llms are not neutral arbiters of information and any llm corporation that's aware the us president is prompting it for answers can ind… | ↗Not briefed |
hn hn-ai | 1mo | How Smart Compilers Unleash GPUs for Scientific Solvers | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | ᴏᴋɪᴇ, ᴛɪᴍᴇ ғᴏʀ ᴍᴇ ᴛᴏ sʜᴀʀᴇ ᴍʏ #ᴘʀᴏᴍᴏsᴋʏ ғᴏʀ ᴏᴄᴛᴏʙᴇʀ ɴᴏᴡ! 🕯️༄˖°.🍂.ೃ࿔*:･🎃 ɪ’ᴍ ᴅᴀᴊᴀʜ ⏾ 𝟸𝟼 ⏾ sʜᴇ/ʜᴇʀ ⏾ ʙɪ 🩷💜💙 (sɪɴɢʟᴇ) ⏾ sᴍᴀʟʟ ʏᴏᴜᴛᴜʙᴇʀ… | ᴏᴋɪᴇ, ᴛɪᴍᴇ ғᴏʀ ᴍᴇ ᴛᴏ sʜᴀʀᴇ ᴍʏ #ᴘʀᴏᴍᴏsᴋʏ ғᴏʀ ᴏᴄᴛᴏʙᴇʀ ɴᴏᴡ! 🕯️༄˖°.🍂.ೃ࿔*:･🎃 ɪ’ᴍ ᴅᴀᴊᴀʜ ⏾ 𝟸𝟼 ⏾ sʜᴇ/ʜᴇʀ ⏾ ʙɪ 🩷💜💙 (sɪɴɢʟᴇ) ⏾ sᴍᴀʟʟ ʏᴏᴜ… | ↗Not briefed |
rss techmeme | 1mo | Lyft agrees to pay $272.5M to settle California claims that it mislabeled drivers as independent contractors rather than employees between 2016 and 2020 (Daniel Wiessner/Reuters) | Daniel Wiessner / Reuters: Lyft agrees to pay $272.5M to settle California claims that it mislabeled drivers as independent contractor… | ↗Not briefed |
hn hn-ai | 1mo | When Fancy Eviction Fails: Rethinking Cache Replacement for LLM Prefix Reuse | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | I keep hearing that AI is going to transform research and publishing. Then in my journal editing role I come across the below - an instru… | I keep hearing that AI is going to transform research and publishing. Then in my journal editing role I come across the below - an i… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Learning the Latent Structure: A Feature-Centric Approach to Graph Data Augmentation | Graph-structured data plays a pivotal role in modeling complex relationships. However, real-world graphs are often incomplete due to d… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Someone once said that Trump's brain tends to work like an LLM which keeps stochastically predicting the next word in sequence without any … | Someone once said that Trump's brain tends to work like an LLM which keeps stochastically predicting the next word in sequence without… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Student-Guided Teacher Distillation for Efficient LLM Task Routing: Positioning Against Jev-Style System-1 Classifiers | Zero-shot classifiers are useful for routing user requests to specialized LLM tasks, but scoring every request against a large candida… | ↗Not briefed |
arxiv arxiv-ai | 1mo | IGNITE Tokamak World Model Architecture | We introduce IGNITE, a generative world foundation model for fusion plasma behavior simulation trained in a self-supervised manner fro… | ↗Not briefed |
arxiv arxiv-ai | 1mo | From Fragments to Global Maps: Learning Vectorized Map Aggregation with Large Language Models | Large-scale vectorized HD maps provide structured road information that is essential for perception, localization, and planning in aut… | ↗Not briefed |
bluesky bluesky-ai | 1mo | “We must be very nice to the AI” is not a fringe view. the people making this stuff are a fucked up cult www.nytimes.com/2026/09/29/u... | “We must be very nice to the AI” is not a fringe view. the people making this stuff are a fucked up cult www.nytimes.com/2026/09/29/u… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Post-Training Quantization of Autoregressive Weather Models | Advancements in high-resolution numerical weather prediction (NWP) and data assimilation (DA) have shaped the developments in deep lea… | ↗Not briefed |
arxiv arxiv-ai | 1mo | On-Premises Multi-Course RAG Tutoring for Business Education: Hardware-Software Trade-offs in a Campus AI Tutor🔧 Ollama | Campus AI tutors based on retrieval-augmented generation (RAG) must ground answers in assigned course materials while keeping textbook… | ↗Not briefed |
bluesky bluesky-ai | 1mo | if you wanted to read hundreds of DDLC pregnancy fanfiction stories then an LLM is really your best option | if you wanted to read hundreds of DDLC pregnancy fanfiction stories then an LLM is really your best option | ↗Not briefed |
arxiv arxiv-ai | 1mo | World Action Modeling with Progressive Visual Planning | World action models (WAMs) have emerged as a promising paradigm for robotic control by jointly predicting future visual dynamics and a… | ↗Not briefed |
bluesky bluesky-ai | 1mo | It's genuinely kind of wild to me how many people are fooled by a senator just delegating all their spoils to their families and extended f… | It's genuinely kind of wild to me how many people are fooled by a senator just delegating all their spoils to their families and exten… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Graphene – Data analysis toolkit for your coding agent | My friend and I have worked at a number of BI companies and thought: Can’t a coding agent do most of this now?<p>Almost. They just nee… | ↗Not briefed |
hn hn-ai | 1mo | The Note – A Markdown notebook with runnable code and formula tables | — | ↗Not briefed |
hn hn-ai | 1mo | Crafting Express – isolated local environments for AI coding agents | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Multi-Fidelity Policy Gradients Stabilize Data-Scarce Reinforcement Learning | Policy gradient methods for on-policy reinforcement learning (RL) can become unstable when expensive, scarce target-domain data yield… | ↗Not briefed |
arxiv arxiv-ai | 1mo | HXAI: Hierarchical Privacy-Preserving Explainable AI in Distributed Energy Systems | Balancing electricity demand and supply is increasingly difficult due to the inherent intermittency of renewable power generation and… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Compound AI System Reliability: A Failure Taxonomy and Resilience Pattern Catalog from 150 Production Incidents | Deploying compound AI systems reliably and safely requires understanding failure modes that emerge at component boundaries, not within… | ↗Not briefed |
bluesky bluesky-ai | 1mo | It's never seen video and it probably isn't really modeling the process of 60's video. Though what's really weird is that this video has ob… | It's never seen video and it probably isn't really modeling the process of 60's video. Though what's really weird is that this video h… | ↗Not briefed |