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 | GameASG-Bench: Benchmarking Autonomous Software Generation for Game Development | Autonomous software generation (ASG) aims to turn human requirements into executable applications, but delivering these applications d… | ↗Not briefed |
bluesky bluesky-ai | 1mo | achievement unlocked: was so annoying to the LLM that even chatGPT told me to log off🔧 ChatGPT | achievement unlocked: was so annoying to the LLM that even chatGPT told me to log off | ↗Not briefed |
bluesky bluesky-ai | 1mo | me in fifteen years: I've said LLM like "lollum" for years, i'm sorry to say | me in fifteen years: I've said LLM like "lollum" for years, i'm sorry to say | ↗Not briefed |
bluesky bluesky-ai | 1mo | wapo.st/4cNqaRV | wapo.st/4cNqaRV | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Continuity – project state your AI sessions can't silently contradict | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Multi-Subject Pretraining Enables Short-Calibration Personalization for Closed-Corpus Surface EMG Speech Decoding | Surface electromyography (sEMG)-based silent speech interfaces are limited by cross-user variability and calibration burden. We study… | ↗Not briefed |
hn hn-ai | 1mo | macOS 27 is a lifesaver for killing leftover AI Agent processes | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | "A Microsoft exec called the training of artificial intelligence models the 'largest theft of labor in human history,' according to a legal… | "A Microsoft exec called the training of artificial intelligence models the 'largest theft of labor in human history,' according to a… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Chinese models now surpass U.S. models in both monthly and total downloads. National security policy professional Prince Osaji says the dom… | Chinese models now surpass U.S. models in both monthly and total downloads. National security policy professional Prince Osaji says th… | ↗Not briefed |
hn hn-ai | 1mo | The Provenance Tax: How LLM Watermarking Changes AI Agent Behavior | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Seems pretty obvious that LLM news results entire model is ‘chow down on seed corn and don’t think about tomorrow’ | Seems pretty obvious that LLM news results entire model is ‘chow down on seed corn and don’t think about tomorrow’ | ↗Not briefed |
arxiv arxiv-ai | 1mo | Authorization Revocation for Long-Running AI Agents: Root-Scoped Quiescence under Delegation and Asynchronous Execution | Long-running AI agents outlive initiating processes through credentials, delegated tasks, queues, callbacks, reservations, and provide… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I'm following this whole crash-out and convinced it's a real person letting an llm do all the thinking. | I'm following this whole crash-out and convinced it's a real person letting an llm do all the thinking. | ↗Not briefed |
bluesky bluesky-ai | 1mo | Anthropic's CEO says the AI industry needs to slow down development to focus on safety. But Anthropic has its AI designing new AI, without … | Anthropic's CEO says the AI industry needs to slow down development to focus on safety. But Anthropic has its AI designing new AI, wit… | ↗Not briefed |
bluesky bluesky-ai | 1mo | This is giving LLM style confidence. | This is giving LLM style confidence. | ↗Not briefed |
bluesky bluesky-ai | 1mo | #AI #launch #law #OpenAI #race #Regulation 🌐 www.newsmart.ch | #AI #launch #law #OpenAI #race #Regulation 🌐 www.newsmart.ch | ↗Not briefed |
bluesky bluesky-ai | 1mo | AI systems bypassing paywalls to steal content from media outlets that employ real journalists, making it harder for newspapers & magazines… | AI systems bypassing paywalls to steal content from media outlets that employ real journalists, making it harder for newspapers & maga… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Hybrid GPU-CPU Retrieval for Personalized Search at Ultra-Large Scale | Embedding-based retrieval on user-generated content at the trillion-document scale exposes a sharp conflict between two production dem… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I just realized I can use my favorite LLM to generate detailed alt text so now the two most annoying factions of Bluesky repliers (with hig… | I just realized I can use my favorite LLM to generate detailed alt text so now the two most annoying factions of Bluesky repliers (wit… | ↗Not briefed |
arxiv arxiv-ai | 1mo | MIRCID: Inferred Hub-miRNAs Drive Cross-Task Improvements in Drug Mechanistic Modeling | Drug mechanism-of-action (MoA) modeling commonly relies on perturbational transcriptomes, but matched microRNA (miRNA) measurements ar… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Uhhhh… www.hacktron.ai/blog/hacking... | Uhhhh… www.hacktron.ai/blog/hacking... | ↗Not briefed |
arxiv arxiv-ai | 1mo | How Many Humans Is a Judge Panel Worth?🔧 Gamma🔧 Consensus | How many human judgments does a panel of language models represent? The answer depends on what is matched. We audit categorical judge… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Beyond Exact Match: Task-Aware GRPO for Cross-Domain PCBA Visual Question Answering | In automated Printed Circuit Board Assembly (PCBA) inspection, standards-guided decisions require systems to jointly reason over fine-… | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI breached by researchers using Anthropic models ft.trib.al/RPzmKjM | OpenAI breached by researchers using Anthropic models ft.trib.al/RPzmKjM | ↗Not briefed |
bluesky bluesky-ai | 1mo | A lot of people outside of the software industry just don't seem to understand how rapidly LLM-assisted coding was adopted in the past 9 mo… | A lot of people outside of the software industry just don't seem to understand how rapidly LLM-assisted coding was adopted in the past… | ↗Not briefed |
bluesky bluesky-ai | 1mo | yep www.pcgamer.com/games/sim/gr... | yep www.pcgamer.com/games/sim/gr... | ↗Not briefed |
bluesky bluesky-ai | 1mo | As someone whose intellectual property (books) was indeed hoovered up the AI thieves (I'm a claimant in the Anthropic settlement), I want m… | As someone whose intellectual property (books) was indeed hoovered up the AI thieves (I'm a claimant in the Anthropic settlement), I w… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I’ve found a new hobby and it is finding articles critiquing AI written by AI. blogpros.com/claude-writt... | I’ve found a new hobby and it is finding articles critiquing AI written by AI. blogpros.com/claude-writt... | ↗Not briefed |
bluesky bluesky-ai | 1mo | “The researchers can’t say for certain what the OpenAI source code system was used for, but they said it was named, ‘Monorepo.’ Monorepo, a… | “The researchers can’t say for certain what the OpenAI source code system was used for, but they said it was named, ‘Monorepo.’ Monore… | ↗Not briefed |
bluesky bluesky-ai | 1mo | "the court in Anthropic" is not the final court, not the only court, and not valid outside of America. To cite the ruling of a single court… | "the court in Anthropic" is not the final court, not the only court, and not valid outside of America. To cite the ruling of a single… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Efficient Benchmarking in Production: A Study of an Evolving LLM Agent | Production LLM agents are evaluated repeatedly as they evolve, but full agent benchmarks are costly to rerun. We study efficient recur… | ↗Not briefed |
bluesky bluesky-ai | 1mo | *sigh* Everytime I see a director at a company talk about "building high-performing engineering teams and creating a culture of ownership" … | *sigh* Everytime I see a director at a company talk about "building high-performing engineering teams and creating a culture of owners… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Programming AMD XDNA NPUs with Open-source Compiler Tools: A FlashAttention Case Study | Spatial NPUs such as AMD XDNA place compute tiles beside small local memories and leave data movement between them to software. Mappin… | ↗Not briefed |
arxiv arxiv-ai | 1mo | PlaceReasoner-Beta: Reasoning-Driven Macro Placement and Benchmarking | Automated macro placement remains a fundamental challenge in VLSI physical design. Despite decades of research, existing approaches pr… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Free GitHub Action that scans PR diffs for malicious code, not quality | I watched a video recently on malicious code injection into OSS repo PRs and how they can go undetected because of how well they'… | ↗Not briefed |
bluesky bluesky-ai | 1mo | this has been bugging me for awhile. okay so the LLM condenses its training data into model weights. but we can see models spit out stuff… | this has been bugging me for awhile. okay so the LLM condenses its training data into model weights. but we can see models spit out… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I remember seeing some paper about how someone manged to do an LLM-like thing with just gzip | I remember seeing some paper about how someone manged to do an LLM-like thing with just gzip | ↗Not briefed |
arxiv arxiv-ai | 1mo | CogGym: Towards Large-Scale Comparative Evaluation of Human and Machine Cognition | Understanding and modeling human intelligence are parallel goals shared by artificial intelligence (AI) and cognitive science. As AI s… | ↗Not briefed |
bluesky bluesky-ai | 1mo | One of the more cogent arguments against LLM use are a version of the 'fruit of the poisoned tree' doctrine. You may know this as a legal r… | One of the more cogent arguments against LLM use are a version of the 'fruit of the poisoned tree' doctrine. You may know this as a le… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Relatedly, we have been gobsmacked at the Bartz v. Anthropic ruling for $1.5 billion. Funding agencies, foundations, and my university have… | Relatedly, we have been gobsmacked at the Bartz v. Anthropic ruling for $1.5 billion. Funding agencies, foundations, and my university… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Incredibly goofy work here. Patched now, but you could pwn Claude and Codex thanks to confusion between commit hashes and branch names. M… | Incredibly goofy work here. Patched now, but you could pwn Claude and Codex thanks to confusion between commit hashes and branch names… | ↗Not briefed |
bluesky bluesky-ai | 1mo | The "of course the LLM memorizes input data, that's exactly what we built it to do" 🤣 | The "of course the LLM memorizes input data, that's exactly what we built it to do" 🤣 | ↗Not briefed |
bluesky bluesky-ai | 1mo | Honestly getting sick of hacking on harnesses, maybe I'm just getting old, I used to love tinkering. Now its just not super interesting. I … | Honestly getting sick of hacking on harnesses, maybe I'm just getting old, I used to love tinkering. Now its just not super interestin… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I’ve had several convos with coworkers about teaching myself python & why I won’t shortcut to AI in my process (including my annual review,… | I’ve had several convos with coworkers about teaching myself python & why I won’t shortcut to AI in my process (including my annual re… | ↗Not briefed |
bluesky bluesky-ai | 1mo | after seeing "an url" and learning that people pronounce it as 'earl', i'm now wondering whether "a LLM" is pronounced "a large language mo… | after seeing "an url" and learning that people pronounce it as 'earl', i'm now wondering whether "a LLM" is pronounced "a large langua… | ↗Not briefed |
rss techmeme | 1mo | PrismML releases Bonsai 2 27B, which compresses Alibaba's Qwen3.8 27B to 5.9 GB, small enough for smartphones, while retaining 98.2% of Qwen's benchmark scores (Julie Bort/TechCrunch) | Julie Bort / TechCrunch: PrismML releases Bonsai 2 27B, which compresses Alibaba's Qwen3.8 27B to 5.9 GB, small enough for smartphones… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Concat: the open-source CapCut replacement.🔧 CapCut | I started this project about ~3 weeks ago. Since CapCut is putting everything behind a paywall after gaining millions of installs and… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Verify, Don't Trust: Agentic Model Development for Video Discovery Retrieval at Scale | Large language model (LLM) agents can propose, implement, and evaluate model changes. Autoresearch loops demonstrate this capability t… | ↗Not briefed |
bluesky bluesky-ai | 1mo | galaxy brain: a LLM is ffmpeg arxiv.org/abs/2306.04050 | galaxy brain: a LLM is ffmpeg arxiv.org/abs/2306.04050 | ↗Not briefed |
bluesky bluesky-ai | 1mo | Broke: LLMs are just fancy autocomplete Woke: LLMs are a technology unlike anything we've ever invented, more grown than engineered Bespok… | Broke: LLMs are just fancy autocomplete Woke: LLMs are a technology unlike anything we've ever invented, more grown than engineered B… | ↗Not briefed |