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 | Timeline-Bench: Evaluating Agents on Realistic Video-Editing Tasks, from Raw Footage to Final Cut | AI agents increasingly carry out long-horizon professional work, but their evaluations rarely require a finished creative deliverable.… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Leí una frase que era: Hemos gastado miles de millones y toneladas de computación en hacer que los ordenadores sean malos en matemáticas (a… | Leí una frase que era: Hemos gastado miles de millones y toneladas de computación en hacer que los ordenadores sean malos en matemátic… | ↗Not briefed |
hn hn-ai | 1mo | Google, OpenAI and Anthropic plan AI safety body, The Information reports | — | ↗Not briefed |
hn hn-ai | 1mo | Cutting LLM tokens on big spreadsheets | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | CacheRepair: Learning to Repair Cross-Chunk Context in RAG for KV Cache Fusion | Multi-document retrieval-augmented generation (RAG) requires a language model to process multiple retrieved text chunks before answeri… | ↗Not briefed |
arxiv arxiv-ai | 1mo | FlexiWorld: Learning and Planning via Flexible Action Chunks Across Multiple Time Scales | Latent world models predict future states for goal-directed planning using action chunks spanning multiple primitive steps. Existing m… | ↗Not briefed |
bluesky bluesky-ai | 1mo | AI-knowers correct me if I’m wrong, but we are getting pretty close to “a lot of people/firms will be able to run a ‘good-enough’ LLM in-ho… | AI-knowers correct me if I’m wrong, but we are getting pretty close to “a lot of people/firms will be able to run a ‘good-enough’ LLM… | ↗Not briefed |
bluesky bluesky-ai | 1mo | yeah... its a shame whenever something new or cool comes out, you gotta second guess or investigate whether AI was involved nowadays before… | yeah... its a shame whenever something new or cool comes out, you gotta second guess or investigate whether AI was involved nowadays b… | ↗Not briefed |
bluesky bluesky-ai | 1mo | For all my whining about the state of "AI" in the open source community, I am glad to see so many OSS projects with firm stances against LL… | For all my whining about the state of "AI" in the open source community, I am glad to see so many OSS projects with firm stances again… | ↗Not briefed |
arxiv arxiv-ai | 1mo | CTP-FL: Common-Trajectory Gradient Prediction for Federated Learning | Communication-efficient federated optimization commonly spends several gradient evaluations between server updates. Existing local-upd… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Explaining Hyperbolic Neural Networks via Geometry-Aware Relevance Propagation | Hyperbolic neural networks introduce geometric operations that require explicit treatment in relevance propagation. Equivalent geometr… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Well, there is something actually, but it doesn't help Anthropic: bsky.app/profile/carl... | Well, there is something actually, but it doesn't help Anthropic: bsky.app/profile/carl... | ↗Not briefed |
hn hn-ai | 1mo | Show HN: OpenAPPA – deterministic AI guardrails that don't break agents | Hi Hacker News! Matvey from Archestra is here.<p>While building enterprise agents, we ran into a problem: the more tools you connect t… | ↗Not briefed |
bluesky bluesky-ai | 1mo | we are in danger of fighting the Last Battle proverbially because LLMs are going to be transformative as media technologies in the same way… | we are in danger of fighting the Last Battle proverbially because LLMs are going to be transformative as media technologies in the sam… | ↗Not briefed |
rss techmeme | 1mo | Cyber defense tech company RedLattice plans to go public via the Bold Eagle Acquisition Corp. SPAC, in a deal valuing RedLattice at ~$1.25B including debt (Liana Baker/Bloomberg) | Liana Baker / Bloomberg: Cyber defense tech company RedLattice plans to go public via the Bold Eagle Acquisition Corp. SPAC, in a deal… | ↗Not briefed |
arxiv arxiv-ai | 1mo | A Multimodal Autonomic Sensing Framework for Objective Assessment of Patient Responses to Dental Pulp Stimulation | Patient responses to dental pulp testing, ranging from no sensation to intense pain, provide important information for assessing pulp… | ↗Not briefed |
hn hn-ai | 1mo | Kleptoparasitism (Food Stealing) in Cetaceans | — | ↗Not briefed |
rss bing-news-ai | 1mo | Contradicting Trump, Pope Leo says artificial intelligence safety concerns not ‘fake news’ | Leo, who dedicated his first encyclical to AI and protecting humanity in its wake, was asked during a press conference about what the… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Tool Mediation Alters Refusal Mechanisms in Large Language Models | Large language models (LLMs) are increasingly deployed with access to external tools, yet harmful tool-mediated interactions are less… | ↗Not briefed |
arxiv arxiv-ai | 1mo | DuplexCadence: Exact State and Execution from a Speech Model's Declared Timelines | Full-duplex speech models support streaming interaction that listens and speaks at the same time. Serving them is governed by a strict… | ↗Not briefed |
arxiv arxiv-ai | 1mo | SymbolicArena: A Unified Infrastructure for Benchmark Distillation and Dynamic Evaluation in Symbolic Regression | Symbolic regression (SR) seeks concise and interpretable mathematical expressions from data for scientific equation discovery. Existin… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Sol-H3: Recursive Self-Improvement for MiniMax-H3 Inference Acceleration on Sol-Engine across Cloud and Edge | Video diffusion models are rapidly scaling and exhibiting enhanced generation capabilities. Among these recent advancements, MiniMax-H… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Jevpipe – a fast System 1 for AI agents, as a Unix pipe | — | ↗Not briefed |
hn hn-ai | 1mo | simdjson 5.0.0 | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Mathematicians and biologists used ChatGPT and Claude for their work only for those companies to later claim AI had discovered solutions to…🔧 ChatGPT | Mathematicians and biologists used ChatGPT and Claude for their work only for those companies to later claim AI had discovered solutio… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Using Context Is Not Enough: Test-Time Training for Personalized Reward Modeling | Reinforcement learning from human feedback (RLHF) aligns large language models (LLMs) with human preferences, yet most pipelines learn… | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI agents hammered a UN data site more than 16,000 times and got around a filter built to block them, according to an independent repor… | OpenAI agents hammered a UN data site more than 16,000 times and got around a filter built to block them, according to an independent… | ↗Not briefed |
bluesky bluesky-ai | 1mo | This? I’m sorry I just am trying to find the whole piece. www.theinformation.com/articles/ant... | This? I’m sorry I just am trying to find the whole piece. www.theinformation.com/articles/ant... | ↗Not briefed |
arxiv arxiv-ai | 1mo | A mechanistic study of language model introspection | Large language models (LLMs) can sometimes report perturbations to their internal activations---even when the input provides no eviden… | ↗Not briefed |
arxiv arxiv-ai | 1mo | DoAtlas-2: A Foundation for Self-Evolving Causal Biomedical Discovery | We introduce DoAtlas-2, a foundation for self-evolving causal biomedical discovery that organizes knowledge around causal mechanisms a… | ↗Not briefed |
bluesky bluesky-ai | 1mo | A 3D shape that can tile 3D space, leaving no gaps and never forming a repeating pattern, has been discovered using an AI model. The resear… | A 3D shape that can tile 3D space, leaving no gaps and never forming a repeating pattern, has been discovered using an AI model. The r… | ↗Not briefed |
hn hn-ai | 1mo | SemIf-OpenJev | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | DRIFT: Disentangled Responsive-Invariant Flow Transport for Single-Cell Perturbation Prediction | Predicting cellular responses to perturbations is a central problem in cellular biology, with broad applications in systems biology an… | ↗Not briefed |
hn hn-ai | 1mo | AI agent that emails, texts, and calls customers – one API, shared context | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | E3J: An Efficient and Open-Source Backend for Euclidean Equivariant Operations on GPU and TPU | We present e3j, a fast Euclid-equivariance backend for geometric deep learning applications with JAX bindings for GPU and TPU. Leverag… | ↗Not briefed |
arxiv arxiv-ai | 1mo | SpikeLite: Lightweight Spiking Neural Networks for Time-Series Forecasting | Spiking neural networks (SNNs) offer an energy-efficient paradigm for time-series forecasting through spike-driven computation. Howeve… | ↗Not briefed |
arxiv arxiv-ai | 1mo | DF-CBM: Region-Aware Concept Bottleneck Models for Deepfake Detection | Deepfake detection methods have become increasingly effective yet most provide limited insight into the evidence behind their predicti… | ↗Not briefed |
hn hn-ai | 1mo | The bgfx shader compiler, 14 years of not being a compiler | — | ↗Not briefed |
hn hn-ai | 1mo | go/codegolf benchmark: frontier models vs. Google's 2015 code golfers | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | As someone that spends way too much time building and using AI, connecting to any kind of OpenAI, Meta (Muse), etc etc agent and giving it … | As someone that spends way too much time building and using AI, connecting to any kind of OpenAI, Meta (Muse), etc etc agent and givin… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Can Generative AI Automate Data Extraction for Meta-Analysis? A Case Study on Intercropping Research | Meta-analysis is the synthesis of information from multiple sources to arrive at an overarching conclusion. There is a large need for… | ↗Not briefed |
bluesky bluesky-ai | 1mo | www.tomshardware.com/tech-industr... | www.tomshardware.com/tech-industr... | ↗Not briefed |
arxiv arxiv-ai | 1mo | When Valid Tool Calls Change Meaning: Formation-Consistent Dispatch for LLM Agents | Tool-enabled agents form calls from model-visible interfaces, while hosts later select their implementation. Standard dispatch omits t… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Retrieval-Augmented Diffusion Modeling for Stochastic Discount Factor Portfolios | In this work, we study portfolio optimization under the stochastic discount factor (SDF) framework by learning market state representa… | ↗Not briefed |
bluesky bluesky-ai | 1mo | "The US came within spitting distance of a war with China over something an LLM made up." -- @mskellymhayes.bsky.social | "The US came within spitting distance of a war with China over something an LLM made up." -- @mskellymhayes.bsky.social | ↗Not briefed |
arxiv arxiv-ai | 1mo | GraphHCA: Closed-Form Hindsight Credit Assignment for Long-Horizon LLM Agents | Group-based reinforcement learning (RL) has advanced large language models (LLMs) and is increasingly extending to agentic tasks, wher… | ↗Not briefed |
rss bing-news-ai | 1mo | Contradicting Trump, Pope Leo says artificial intelligence safety concerns not ‘fake news’ | Pope Leo XIV says concerns about artificial intelligence going rogue are not “fake news” and should be taken seriously. | ↗Not briefed |
rss bing-news-ai | 1mo | Contradicting Trump, Pope Leo says artificial intelligence safety concerns not ‘fake news’ | Pope Leo XIV says concerns about artificial intelligence going rogue are not “fake news” and should be taken seriously. | ↗Not briefed |
rss bing-news-ai | 1mo | Contradicting Trump, Pope Leo says artificial intelligence safety concerns not ‘fake news’ | Pope Leo XIV says concerns about artificial intelligence going rogue are not “fake news” and should be taken seriously. | ↗Not briefed |
rss bing-news-ai | 1mo | Contradicting Trump, Pope Leo says artificial intelligence safety concerns not ‘fake news’ | Pope Leo XIV says concerns about artificial intelligence going rogue are not “fake news” and should be taken seriously. | ↗Not briefed |