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 | 26d | Training Reasoning Models to Generate Diverse Scientific Ideas | One of the first steps of a research pipeline is generating ideas. Large Language Models (LLMs) are increasingly used to automate this… | ↗Not briefed |
arxiv arxiv-ai | 26d | Learning Self-Orchestration in Decentralized Agent Swarms | The capabilities of an agent swarm depend on both its individual members and how they coordinate. Yet existing multi-agent systems oft… | ↗Not briefed |
arxiv arxiv-ai | 26d | AVC-Reward: Event-Centric Audio-Video Consistency Evaluation🔧 Modal | As joint audio-video generation models become more capable, they can generate audio-visual scenes exhibiting three forms of complexity… | ↗Not briefed |
bluesky bluesky-ai | 26d | Honestly the thing that makes me wryly laugh is people assuming AI companies failing means those services don't go open source. You do und… | Honestly the thing that makes me wryly laugh is people assuming AI companies failing means those services don't go open source. You d… | ↗Not briefed |
arxiv arxiv-ai | 26d | A team of Gemini 4 Argon agents beats Minecraft | We placed four Gemini 4 Argon agents in a sandboxed Minecraft survival world and gave them one goal: defeat the Ender Dragon. Each age… | ↗Not briefed |
arxiv arxiv-ai | 26d | Learning to Steer LLM Agents and Multi-Agent Systems | Large language model (LLM)-based agents and multi-agent systems increasingly tackle complex tasks through long-horizon tool use and in… | ↗Not briefed |
arxiv arxiv-ai | 26d | MindEval2: Clinically Grounded Patients and Rubrics for Evaluating Language Models in Mental Health Support | With people increasingly turning to AI chatbots for mental health support, the need to evaluate these tools is becoming more pressing.… | ↗Not briefed |
arxiv arxiv-ai | 26d | Grid-Bench: Can LLMs Predict How Discrete Worlds Change? | Spatial instructions describe how to move objects, transform regions, or modify their surroundings. When several operations are applie… | ↗Not briefed |
rss techmeme | 26d | Memo: Accenture has told many employees, some of whom had been spending as much as $1,000 monthly, that it will now only spend $29 a month per person on Claude (Joseph Cox/404 Media) | Joseph Cox / 404 Media: Memo: Accenture has told many employees, some of whom had been spending as much as $1,000 monthly, that it wil… | ↗Not briefed |
bluesky bluesky-ai | 26d | LLM should stand for lovely large man. | LLM should stand for lovely large man. | ↗Not briefed |
bluesky bluesky-ai | 26d | like, sorry, but I don't want to debate with people claiming their LLM has full agency and consciousness it just makes me really fucking s… | like, sorry, but I don't want to debate with people claiming their LLM has full agency and consciousness it just makes me really fuck… | ↗Not briefed |
arxiv arxiv-ai | 26d | Knowledge That Is Linear Before Fine-Tuning Generalizes After It | Not all facts a language model can recall support further learning equally well. We distinguish active facts, which support generaliza… | ↗Not briefed |
arxiv arxiv-ai | 26d | Testing the Planning Claims of LLM-Based Agents against Formal Semantics | LLM-based agents are increasingly credited with the competences of practical reasoning: planning, monitoring, and correcting their own… | ↗Not briefed |
arxiv arxiv-ai | 26d | AI Consilium: a Benchmark of LLM Clinical Multidisciplinary Second Opinions🔧 Consensus🔧 GPT-5 | Large language models (LLMs) can generate differential diagnoses for complex cases, but a useful second opinion must also be factually… | ↗Not briefed |
rss techmeme | 26d | Anthropic says it is giving employee-like access to METR as part of its pledge to partner with evaluators for assessment of its models and safety practices (Anthropic) | Anthropic: Anthropic says it is giving employee-like access to METR as part of its pledge to partner with evaluators for assessment of… | ↗Not briefed |
bluesky bluesky-ai | 26d | You ever stop to consider that the state of your soul is impacting the state of the world you live in? Like you using an LLM to ask if your… | You ever stop to consider that the state of your soul is impacting the state of the world you live in? Like you using an LLM to ask if… | ↗Not briefed |
arxiv arxiv-ai | 26d | Private Retrieval Augmented Generation via Random Projection | Retrieval-Augmented Generation (RAG) enhances the capabilities of large language models (LLMs) by querying external structured knowled… | ↗Not briefed |
arxiv arxiv-ai | 26d | Backward Compatibility in Zero-Bit Neural Audio Watermarking | Neural watermarking systems are typically trained as tightly coupled embedder-detector pairs and implicitly assume static deployment.… | ↗Not briefed |
arxiv arxiv-ai | 26d | Multi-Target Few-Shot Prompting with Diffusion Language Models | Few-shot prompting is useful when instructions and rewards are hard to specify. One such application is editing by example, where a de… | ↗Not briefed |
arxiv arxiv-ai | 26d | Towards Continual Learning: Mitigating Forgetting With Subspace Tuning | Large language models (LLMs) acquire new skills through fine-tuning, but it can destroy capabilities they already have: catastrophic f… | ↗Not briefed |
bluesky bluesky-ai | 26d | Go on then @societyofauthors.bsky.social , be transparent about who's using AI in your organisation right now, which model and for what pur… | Go on then @societyofauthors.bsky.social , be transparent about who's using AI in your organisation right now, which model and for wha… | ↗Not briefed |
arxiv arxiv-ai | 26d | From Parallels to Divergence: Language Models' Learning of Action Nominal Constructions | Action nominal constructions (ANCs) express events in nominal form, with their argument marking following patterns found in clauses or… | ↗Not briefed |
arxiv arxiv-ai | 26d | CAT-alyzing Honesty: Continuous Adversarial Training for Transferable LLM Honesty | Large language models (LLMs) are becoming prevalent in scenarios where honesty is critical. However, despite alignment training, these… | ↗Not briefed |
arxiv arxiv-ai | 26d | EliteReplay: Preserving Diverse Solutions in RL Post-Training via Quality-Diversity Search | Reinforcement learning has become a common post-training technique to improve the accuracy of LLMs on verifiable tasks. However, such… | ↗Not briefed |
arxiv arxiv-ai | 26d | Improving Environment-Agnostic Coding Agents: The Role of Scaffolding🔧 GPT-5 | Stronger language models may increasingly construct useful procedures for themselves, reducing the advantage of developer-supplied sca… | ↗Not briefed |
arxiv arxiv-ai | 26d | Learning Rational Strategies for Temporal Objectives in Multi-Agent Systems | We study rational synthesis for stochastic multi-agent systems with unknown dynamics, where each agent is equipped with its own probab… | ↗Not briefed |
bluesky bluesky-ai | 26d | In a new essay, David Hall @dlwh.bsky.social argues that pretraining consumes documents, while posttraining consumes tasks. Open language… | In a new essay, David Hall @dlwh.bsky.social argues that pretraining consumes documents, while posttraining consumes tasks. Open lan… | ↗Not briefed |
hn hn-ai | 26d | Linus Torvalds on AI contribution rules | — | ↗Not briefed |
bluesky bluesky-ai | 26d | New: Accenture has told employees it will only fund $29 per month of Claude use. Meaning that employees who often spend as much as $1,000 a… | New: Accenture has told employees it will only fund $29 per month of Claude use. Meaning that employees who often spend as much as $1,… | ↗Not briefed |
hn hn-ai | 26d | Show HN: Voice Standard Bench – Open benchmark of LLMs on speech transcripts | — | ↗Not briefed |
arxiv arxiv-ai | 26d | Asking Smarter Questions: Targeted Active Preference Learning for Underspecified Rewards | Reward learning typically assumes the human's input covers everything that matters for the task. In practice, humans can only attend t… | ↗Not briefed |
hn hn-ai | 26d | Show HN: Blurt – Open-source macOS dictation app (Swift, MIT) | — | ↗Not briefed |
bluesky bluesky-ai | 26d | Anthropic should delay its US IPO until it addresses concerns about safety, governance and the pace of development, urged New York City Com… | Anthropic should delay its US IPO until it addresses concerns about safety, governance and the pace of development, urged New York Cit… | ↗Not briefed |
arxiv arxiv-ai | 26d | LEASH: Uncertainty-Aware Weak-to-Strong Escalation and Oversight for Computer-Use Agents | Computer-use agents increasingly tackle long-horizon tasks, where running a frontier model at every step is expensive and a small open… | ↗Not briefed |
bluesky bluesky-ai | 26d | www.joemygod.com/2026/10/eigh... | www.joemygod.com/2026/10/eigh... | ↗Not briefed |
arxiv arxiv-ai | 26d | Progressive Misalignment Can Emerge from Benign Recursive Self-Distillation | Training models on their own outputs has attracted widespread attention as a possible route to advance AI capabilities. One prominent… | ↗Not briefed |
arxiv arxiv-ai | 26d | Harness-Delta Attribution: Dissecting Gains From Harness Evolution | Harness evolution automatically optimizes the prompts, skills, tools, and orchestration code around a fixed-weight model, offering a w… | ↗Not briefed |
hn hn-ai | 26d | PCEI – On-device embodied AI combining BitNet 1.58B, Mamba, and FastKAN | — | ↗Not briefed |
bluesky bluesky-ai | 26d | My dog is more of a person than your LLM. | My dog is more of a person than your LLM. | ↗Not briefed |
bluesky bluesky-ai | 26d | As indebted AI companies struggle to build up revenue and their charging model is expensive tokens expect more of this. They’re cooked | As indebted AI companies struggle to build up revenue and their charging model is expensive tokens expect more of this. They’re cooked | ↗Not briefed |
bluesky bluesky-ai | 26d | "For a sample of 100 fact-checked posts, 54.5% of Grok bot ratings and 57.7% of Perplexity bot ratings agreed with ratings of human fact-ch… | "For a sample of 100 fact-checked posts, 54.5% of Grok bot ratings and 57.7% of Perplexity bot ratings agreed with ratings of human fa… | ↗Not briefed |
bluesky bluesky-ai | 26d | We’ve committed to having evaluators assess our safety practices. We're in discussions with multiple organizations, and have now given empl… | We’ve committed to having evaluators assess our safety practices. We're in discussions with multiple organizations, and have now given… | ↗Not briefed |
bluesky bluesky-ai | 26d | several top researchers at openAI have started smoking because they are confident my dick and balls will cure cancer | several top researchers at openAI have started smoking because they are confident my dick and balls will cure cancer | ↗Not briefed |
bluesky bluesky-ai | 26d | Recent analysis suggests that as some of the leading AI image generators from Google and OpenAI become more capable, they may be getting wo… | Recent analysis suggests that as some of the leading AI image generators from Google and OpenAI become more capable, they may be getti… | ↗Not briefed |
rss techmeme | 26d | Sunbound, which makes software for handling billing, insurance claims, and payments for senior living operators, raised a $24M Series A led by TTV Capital (Ryan Lawler/Axios) | Ryan Lawler / Axios: Sunbound, which makes software for handling billing, insurance claims, and payments for senior living operators,… | ↗Not briefed |
arxiv arxiv-ai | 26d | Librarian Agents: Training LLMs to Curate Long-Context Workspaces that Anticipate Future Queries | Reasoning over long document contexts remains a challenge for large language models (LLMs). Standard retrieval methods are inherently… | ↗Not briefed |
arxiv arxiv-ai | 26d | X-RIC: An Agentic RAN-Intelligent Controller Benchmark for Enhancing Reinforcement Learning | The O-RAN ALLIANCE introduces architectural flexibility through disaggregation and AI-driven control. While the RAN Intelligent Contro… | ↗Not briefed |
bluesky bluesky-ai | 26d | I feel like people want perfect solutions when it comes to LLM use in fiction and none exist, so they get mad if a magazine accidentally pu… | I feel like people want perfect solutions when it comes to LLM use in fiction and none exist, so they get mad if a magazine accidental… | ↗Not briefed |
hn hn-ai | 26d | Anthropic Update on Independent Evaluators | — | ↗Not briefed |
bluesky bluesky-ai | 26d | Why are eight US states suing OpenAI? https://aje.news/t593mj | Why are eight US states suing OpenAI? https://aje.news/t593mj | ↗Not briefed |