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 | 14d | From Reasoning to Action: Benchmarking Frontier AI Models on LIBERO-10 Robot Manipulation Tasks | Recent advances in frontier AI models have demonstrated remarkable capabilities in visual understanding, reasoning, planning, and tool… | ↗Not briefed |
arxiv arxiv-ai | 14d | How Prompt Style Affects AI Accuracy and Speed in Competitive Programming🔧 GPT-5 | Large language models are increasingly used in competitive programming, and their outputs are sensitive to how a prompt is phrased. Pr… | ↗Not briefed |
bluesky bluesky-ai | 14d | I would be more inclined to withhold my anger if I wasn't dead certain that the only conversation inside OpenAI for the last 36 hours was "… | I would be more inclined to withhold my anger if I wasn't dead certain that the only conversation inside OpenAI for the last 36 hours… | ↗Not briefed |
bluesky bluesky-ai | 14d | Vulnerability research is something that LLM/GenAI is genuinely extremely useful for so of course OpenAI fucks it up because they're so des… | Vulnerability research is something that LLM/GenAI is genuinely extremely useful for so of course OpenAI fucks it up because they're s… | ↗Not briefed |
rss the-decoder | 14d | Deepmind founder Hassabis shortens AGI timeline from five to two to three years | Demis Hassabis keeps moving up his AGI timeline. The Nobel laureate and former Google Deepmind CEO now expects artificial general inte… | ↗Not briefed |
bluesky bluesky-ai | 14d | LLM based coding agents are now a real force. They have a longer history (like all tech) than most people believe. I was at a fireside chat… | LLM based coding agents are now a real force. They have a longer history (like all tech) than most people believe. I was at a fireside… | ↗Not briefed |
bluesky bluesky-ai | 14d | Say what you want about OpenAI | Say what you want about OpenAI | ↗Not briefed |
arxiv arxiv-ai | 14d | SWE-DIVE: Multi-Turn Coding Agent Evaluation with Dynamic User Simulation | Coding agents are typically evaluated using one-shot, well-formulated prompts. Benchmarks such as SWE-bench use existing GitHub issues… | ↗Not briefed |
bluesky bluesky-ai | 14d | Industries tend to pursue self-regulation when they're facing a potential regulatory threat and/or consumer backlash It's a dog and pony s… | Industries tend to pursue self-regulation when they're facing a potential regulatory threat and/or consumer backlash It's a dog and p… | ↗Not briefed |
hn hn-ai | 14d | M5 Stack Cardputer MiniShell | — | ↗Not briefed |
rss techmeme | 14d | The PHP Foundation says its RCE patches are limited to PHP 8.2 and newer, leaving legacy deployments dependent on third-party vendors (Wired) | Wired: The PHP Foundation says its RCE patches are limited to PHP 8.2 and newer, leaving legacy deployments dependent on third-party v… | ↗Not briefed |
bluesky bluesky-ai | 14d | I do think it’s on OpenAI not to brag about their plans going well when that is clearly not what happened. Be transparent about it. PHP was | I do think it’s on OpenAI not to brag about their plans going well when that is clearly not what happened. Be transparent about it. PH… | ↗Not briefed |
arxiv arxiv-ai | 14d | Which Differences Matter? Controlled Tests of Rule Transfer, Joint Reliability, and Information Use in Decision Models | Decision models map stated rules and observed states to discrete answers. Aggregate accuracy leaves unresolved which rule and informat… | ↗Not briefed |
arxiv arxiv-ai | 14d | SP-HLM: Sparse-Pose-Driven Motion Generation and Whole-Body Control for Humanoid Loco-Manipulation | Humanoid loco-manipulation requires an effective interface between high-level task policies and whole-body control. Existing interface… | ↗Not briefed |
arxiv arxiv-ai | 14d | HGM-Dual: Lineage-Aware Guided Agent Harness Optimization for Long-Horizon Planning | Harness optimization, which improves an agent's prompts, tools, and workflows, is an increasingly effective way to enhance agents buil… | ↗Not briefed |
arxiv arxiv-ai | 14d | Robust Soft Skill Adaptation via Behavioral Localization in Tool-Using Agents | Text skills can extend the capabilities of tool-using agents, but long skill descriptions increase computational costs. Soft prompts p… | ↗Not briefed |
bluesky bluesky-ai | 14d | Increíble la cantidad de AI Slop que consumen las viejas. novelas de facebook, posteos de "si querés que me salve poné AMÉN", recetas de co… | Increíble la cantidad de AI Slop que consumen las viejas. novelas de facebook, posteos de "si querés que me salve poné AMÉN", recetas… | ↗Not briefed |
rss openai-news | 14d | How AT&T is rethinking in-house legal work with OpenAI | AT&T is working with OpenAI to make legal knowledge easier to access. Learn how its LegalEdge team is rethinking intake and scaling le… | ↗Not briefed |
bluesky bluesky-ai | 14d | This might explain some of it. Does Heather Cox Richardson know that Tester is on the board of Anthropic? Or that he bullied the Vice Chair… | This might explain some of it. Does Heather Cox Richardson know that Tester is on the board of Anthropic? Or that he bullied the Vice… | ↗Not briefed |
bluesky bluesky-ai | 14d | Like, uh... afaict there is this narrative that OpenAI is somehow the actor that did early disclosure for the PHP vuln. But every credible … | Like, uh... afaict there is this narrative that OpenAI is somehow the actor that did early disclosure for the PHP vuln. But every cred… | ↗Not briefed |
hn hn-ai | 14d | Ask HN: Financial Audit of DuckDuckGo? | Duckduckgo has really expanded their offerings, including AI search results. This seems expensive. It makes me suspicious that it coul… | ↗Not briefed |
bluesky bluesky-ai | 14d | Si OpenAI decide no salir a bolsa es la prueba de que ya no engañan a nadie, pero si Anthropic sale a bolsa es la prueba de que la bolsa es… | Si OpenAI decide no salir a bolsa es la prueba de que ya no engañan a nadie, pero si Anthropic sale a bolsa es la prueba de que la bol… | ↗Not briefed |
hn hn-ai | 14d | A new Speculative decoding drafter for vLLM and SGLang called EAGLE-4🔧 vLLM | — | ↗Not briefed |
bluesky bluesky-ai | 14d | Et je ne parle même pas de tout le gloubiboulga infâme autour de la publication des détails de ces vulnérabilités. Plusieurs acteurs commer… | Et je ne parle même pas de tout le gloubiboulga infâme autour de la publication des détails de ces vulnérabilités. Plusieurs acteurs c… | ↗Not briefed |
bluesky bluesky-ai | 14d | You should never, ever trust an LLM - the "problem" of hallucinations is fundamental to the way they work, and even the big AI companies ha… | You should never, ever trust an LLM - the "problem" of hallucinations is fundamental to the way they work, and even the big AI compani… | ↗Not briefed |
hn hn-ai | 14d | Php2shell-Checker | — | ↗Not briefed |
bluesky bluesky-ai | 14d | LLM outputs are, shall we say, *inconsistent* at the best of times. Sometimes the output it good, sometimes the output is bad. When the out… | LLM outputs are, shall we say, *inconsistent* at the best of times. Sometimes the output it good, sometimes the output is bad. When th… | ↗Not briefed |
bluesky bluesky-ai | 14d | Canadian publishers Postmedia, Torstar, The Globe and Mail and The Canadian Press have filed a copyright lawsuit against Anthropic. They a… | Canadian publishers Postmedia, Torstar, The Globe and Mail and The Canadian Press have filed a copyright lawsuit against Anthropic. T… | ↗Not briefed |
arxiv arxiv-ai | 14d | What the Author Wrote, What the Agent Reads: Silent Cue Loss in Enterprise Document Ingestion🔧 GPT-5 | Enterprise agents act on documents. Before an agent can follow a policy, answer a compliance question or file a ticket, some ingestion… | ↗Not briefed |
bluesky bluesky-ai | 14d | Any organisation applying LLM-generated changes directly to code that the Internet runs on is about to find out why lawyers are so expensiv… | Any organisation applying LLM-generated changes directly to code that the Internet runs on is about to find out why lawyers are so exp… | ↗Not briefed |
bluesky bluesky-ai | 14d | Had a nightmare Sony contracted me to playtest LBP4 in private development, they introduced a generative AI model to the create mode that u… | Had a nightmare Sony contracted me to playtest LBP4 in private development, they introduced a generative AI model to the create mode t… | ↗Not briefed |
arxiv arxiv-ai | 14d | Stationarity Constraints Improve Both Energies and Densities of Machine-Learned Exchange-Correlation Functionals | Empirical exchange-correlation functionals are fitted on fixed orbitals but used self-consistently, so better energies often come with… | ↗Not briefed |
bluesky bluesky-ai | 14d | Using the most “advanced” tech as a metaphor for brains while the most “advanced” tech is trying to emulate brains, but faster, feels like … | Using the most “advanced” tech as a metaphor for brains while the most “advanced” tech is trying to emulate brains, but faster, feels… | ↗Not briefed |
bluesky bluesky-ai | 14d | OpenAI worked with a group of several security companies and maintainers to find the vulnerabilities and one security company who has a his… | OpenAI worked with a group of several security companies and maintainers to find the vulnerabilities and one security company who has… | ↗Not briefed |
bluesky bluesky-ai | 14d | www.reddit.com/r/technology... | www.reddit.com/r/technology... | ↗Not briefed |
arxiv arxiv-ai | 14d | ExGrad: Beyond Suppression - Explicit Clean-Gradient Subtraction in Adversarial Training | Adversarial training derives updates from surrogate losses, but source-loss descent does not directly characterize an update's effect… | ↗Not briefed |
hn hn-ai | 14d | Show HN: IDE for a general-purpose visual programming language "Pipe" | I'm happy to announce the release of the web-based Pipe IDE, availablle by clicking button "Try Pipe IDE" on website.<p… | ↗Not briefed |
arxiv arxiv-ai | 14d | GuideEvolve: Self-Evolving Agent Framework with Expert Guidance and 3Ts Mechanisms | This study introduces GuideEvolve, a self-evolving agent framework incorporating expert guidance and three autonomous regulatory mecha… | ↗Not briefed |
bluesky bluesky-ai | 14d | LLMと画像や動画の生成モデルがまとめてAIって呼ばれてるせいで結構ムズムズする会話があちこちで繰り広げられてる🥺 例えばだけどLLM悪用によるハッキングと画像動画生成モデルへの創作物の無断学習ってあんま関係ないでしょ🤔 | LLMと画像や動画の生成モデルがまとめてAIって呼ばれてるせいで結構ムズムズする会話があちこちで繰り広げられてる🥺 例えばだけどLLM悪用によるハッキングと画像動画生成モデルへの創作物の無断学習ってあんま関係ないでしょ🤔 | ↗Not briefed |
arxiv arxiv-ai | 14d | Code Comprehensor: Hierarchical Documentation and Cross-Modal Retrieval for Large Repositories🔧 Modal | Understanding large, complex, and rapidly evolving code bases remains a core challenge for software engineering at scale, hindering on… | ↗Not briefed |
hn hn-ai | 14d | Physicist Christopher Stubbs Joins Anthropic's Long-Term Benefit Trust | — | ↗Not briefed |
bluesky bluesky-ai | 14d | imagine if you could only write one word at a time no takes-backsies honestly I kinda wonder what an LLM with a "that was stupid, delete l… | imagine if you could only write one word at a time no takes-backsies honestly I kinda wonder what an LLM with a "that was stupid, del… | ↗Not briefed |
hn hn-ai | 14d | Expert Warned 'Always Verify' AI, Then Didn't Verify His AI-Made Charts | — | ↗Not briefed |
arxiv arxiv-ai | 14d | Agent Behavior Maps: Locating Agents by What They Would Do, Without Running Them | LLM agents with similar success rates can follow different action sequences. An Agent Behavior Map represents LLM-harness combinations… | ↗Not briefed |
hn hn-ai | 14d | Nvidia GPUs sit idle in Meta's Prometheus | — | ↗Not briefed |
rss bing-news-ai | 14d | How artificial intelligence is helping doctors and patients fight breast cancer | Artificial intelligence is helping make radiation more precise for patients with breast cancer. Health reporter Stephanie Stahl has th… | ↗Not briefed |
bluesky bluesky-ai | 14d | Yeah, I have to patch an average of a dozen libraries every week at work because of critical or high vulnerabilities, something that didn't… | Yeah, I have to patch an average of a dozen libraries every week at work because of critical or high vulnerabilities, something that d… | ↗Not briefed |
arxiv arxiv-ai | 14d | Mitigating Model Collapse through Dreaming Learning | Training generative models on synthetic data causes loss of diversity and a decrease in performance, a phenomenon known as model colla… | ↗Not briefed |
bluesky bluesky-ai | 14d | Meanwhile my dumbass has been trying to resurrect a dead Darktide mod for my own super niche use by reading the modding tools discord stuff… | Meanwhile my dumbass has been trying to resurrect a dead Darktide mod for my own super niche use by reading the modding tools discord… | ↗Not briefed |
arxiv arxiv-ai | 14d | Building Auditable AI Text Detectors with Sparse Autoencoders | AI-generated text detectors are increasingly deployed in consequential settings---peer review, education, admissions, journalism, and… | ↗Not briefed |