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 |
|---|---|---|---|---|
bluesky bluesky-ai | 8d | selfpublishingadvice.org/collins-crow... "Mogging" may have won, but the runners-up are excellent. If I could have voted, I would have g… | selfpublishingadvice.org/collins-crow... "Mogging" may have won, but the runners-up are excellent. If I could have voted, I would h… | ↗Not briefed |
arxiv arxiv-ai | 8d | Bellman Contraction under MMD: A Unified Framework | We introduce a mathematical framework for studying contraction of the distributional Bellman operator under Maximum Mean Discrepancy (… | ↗Not briefed |
bluesky bluesky-ai | 8d | LMFAO ANTHROPIC FILED THE IPO SAME MORNING JENSEN GETS FIRED FUCK IT WE BALL | LMFAO ANTHROPIC FILED THE IPO SAME MORNING JENSEN GETS FIRED FUCK IT WE BALL | ↗Not briefed |
bluesky bluesky-ai | 8d | A model backdoor embeds hidden behavior that stays dormant until a trigger appears. In an agent, it can invoke tools, execute commands, or … | A model backdoor embeds hidden behavior that stays dormant until a trigger appears. In an agent, it can invoke tools, execute commands… | ↗Not briefed |
hn hn-ai | 8d | OpenIGI: WebGPU independent reimplementation of Project IGI | — | ↗Not briefed |
bluesky bluesky-ai | 8d | this is the real life version of the LLM memes where it's like "I see now that I lacked decorum and disrespected fallen soldiers. Thank you… | this is the real life version of the LLM memes where it's like "I see now that I lacked decorum and disrespected fallen soldiers. Than… | ↗Not briefed |
arxiv arxiv-ai | 8d | CapLite: Prompt-Injection Defense for Small Language Model Agents without Model-Written Security Artifacts | Enforcement defenses against prompt injection, such as CaMeL, Progent, IPIGuard and DRIFT, ask the agent's own model to write a securi… | ↗Not briefed |
arxiv arxiv-ai | 8d | Designing Institutions for Autonomous AI | As autonomous AI agents increasingly interact in persistent populations, their collective capabilities and alignment depend not only o… | ↗Not briefed |
bluesky bluesky-ai | 8d | Is there an LLM you can practice with? | Is there an LLM you can practice with? | ↗Not briefed |
hn hn-ai | 8d | I made my GitHub contribution graph say "POOP" | — | ↗Not briefed |
arxiv arxiv-ai | 8d | SALA: Semantically Aligned Latent Actions Enable Dual-Axis Scaling of Embodied Foundation Models | Embodied foundation models hinge on diverse robot interaction data, which are costly to collect. Egocentric human videos offer a compl… | ↗Not briefed |
arxiv arxiv-ai | 8d | Near-Optimal Regret for Oracle-Efficient Adversarial Contextual Bandits | We study contextual bandits with $K$ actions, a class of $N$ policies, and i.i.d.\ contexts from a distribution that the learner can s… | ↗Not briefed |
hn hn-ai | 8d | Show HN: GraphsWithStraightLines.com – 126 trends extended to 2050 | I got curious about what we can and can't learn about the future by plotting straight lines on graphs so I made a little visualis… | ↗Not briefed |
arxiv arxiv-ai | 8d | LazyWeights: Decode-on-Demand Edge LLMs | Memory decides which LLMs an edge device runs, and how fast: capacity on a $4$ GB Jetson Nano, bandwidth on an $8$ GB Jetson Orin Nano… | ↗Not briefed |
arxiv arxiv-ai | 8d | Disentangling Spurious Correlations: How IRM Handles Diverse Shortcuts in Text Classification | In NLP, models trained via Empirical Risk Minimization (ERM) often exploit spurious correlations, such as dialectal markers, at the ex… | ↗Not briefed |
hn hn-ai | 8d | Show HN: My $300 embroidery machine needed a subscription, so I built my own app | — | ↗Not briefed |
arxiv arxiv-ai | 8d | Predicting Ethanol Drinking Status from Nonhuman Primate Liver Histology: A Proof of Concept for Parameter-efficient Fine-tuning in Few-Shot Vision-Language Modeling | Combining vision-language models (VLMs) with parameter-efficient fine-tuning (PEFT) offers a promising approach to few-shot whole-slid… | ↗Not briefed |
bluesky bluesky-ai | 8d | Hearing someone asked AI to restart a server and it ended up destroying it made me smile. github.com/openai/codex... | Hearing someone asked AI to restart a server and it ended up destroying it made me smile. github.com/openai/codex... | ↗Not briefed |
bluesky bluesky-ai | 8d | Es erscheint mir nicht so aber im Zeitalter von LLM, wer kann es sagen. | Es erscheint mir nicht so aber im Zeitalter von LLM, wer kann es sagen. | ↗Not briefed |
arxiv arxiv-ai | 8d | Joint Null-Space Projection for Data-Free Model Merging🔧 Consensus | Merging independently fine-tuned checkpoints into a single multi-task model requires no data, no labels, and no gradient steps. In pra… | ↗Not briefed |
hn hn-ai | 8d | Anthropic's Model Taught Itself to Run on AMD in "Three Days." | — | ↗Not briefed |
rss techmeme | 8d | Samsung's Android XR smart glasses hands-on: stylish with good privacy protections like local face redaction and nice Gemini features like real-time translation (Victoria Song/The Verge) | Victoria Song / The Verge: Samsung's Android XR smart glasses hands-on: stylish with good privacy protections like local face redactio… | ↗Not briefed |
rss techcrunch-ai | 8d | SoftBank’s billion-dollar loss may have put a cap on AI’s natural gas ambitions | SoftBank lost over $3 billion pre-ordering natural gas turbines to power AI data centers. Here's what the loss means to power markets. | ↗Not briefed |
hn hn-ai | 8d | How much of GitHub's most-starred AI tooling is written by coding agents? | — | ↗Not briefed |
arxiv arxiv-ai | 8d | A Compact xDAWN-Guided Spatio-Temporal Network with Filter-Bank Convolution for P300 Detection | P300-based brain-computer interfaces (BCIs) enable communication by decoding stimulus-evoked EEG responses into character selections w… | ↗Not briefed |
rss theverge-ai | 8d | More game developers are using generative AI than they did last year | A photo from the GDC Festival of Gaming 2026 event. More video game developers are using generative AI tools in their jobs compared… | ↗Not briefed |
rss techcrunch-ai | 8d | Phota Labs uses AI to turn everyday photos into professional-looking shots | Rather than giving consumers yet another photo-editing tool, Phota Labs aims to use advances in AI to automate the editing process so… | ↗Not briefed |
rss theverge-ai | 8d | AI chatbots are recommending fake retail websites, study finds🔧 ChatGPT | Shoppers using AI chatbots to research products or find places to buy items could be directed to imposters and other risky websites, a… | ↗Not briefed |
hn hn-ai | 8d | What Is Culture? | — | ↗Not briefed |
hn hn-ai | 8d | AWS Lambda Web Functions keep billing memory ~7 minutes after the last request | — | ↗Not briefed |
arxiv arxiv-ai | 8d | Speed-to-Power Dominates Energy Costs for AI Data Centers: Technoeconomic Analysis of Powering AI in the Continental United States | The rapid growth of artificial intelligence is driving demand for new data centers at large scale and on short deployment timelines. T… | ↗Not briefed |
bluesky bluesky-ai | 8d | This is still burning my ass, so next time I'm going to ask how much of my assessment is being offloaded to an LLM and how much money that … | This is still burning my ass, so next time I'm going to ask how much of my assessment is being offloaded to an LLM and how much money… | ↗Not briefed |
rss the-decoder | 8d | OpenAI’s data suggests that frontier models are hiding their thoughts because developers optimize for capability and compute | OpenAI's safety team systematically investigated why the chains of thought in frontier models like GPT-6 Astra are increasingly evadin… | ↗Not briefed |
bluesky bluesky-ai | 8d | Few things move my internal needle toward “AI hater” more than being given some LLM slop masquerading as a brief. Argh. | Few things move my internal needle toward “AI hater” more than being given some LLM slop masquerading as a brief. Argh. | ↗Not briefed |
bluesky bluesky-ai | 8d | Google Chrome is worse than malware at this point, delete it and change your browser for yesterday if you want a computer that's actually y… | Google Chrome is worse than malware at this point, delete it and change your browser for yesterday if you want a computer that's actua… | ↗Not briefed |
arxiv arxiv-ai | 8d | Optimal Single-History Tokenization Disambiguation for Provably Secure Linguistic Steganography | Provably secure linguistic steganography assumes that the receiver recovers the sender's tokens, but a text channel delivers only text… | ↗Not briefed |
bluesky bluesky-ai | 8d | Hey just as a reminder, every current useful large language models was built on the theft of other people's labor. Nobody was paid for th… | Hey just as a reminder, every current useful large language models was built on the theft of other people's labor. Nobody was paid f… | ↗Not briefed |
hn hn-ai | 8d | Compress any LLM 3x in one click, near-lossless, vLLM successor🔧 vLLM | — | ↗Not briefed |
bluesky bluesky-ai | 8d | Sobre buscas tradicionais x IAs generativas e uso de textos de livros piratas: "A busca copiava o conteúdo e devolvia leitores. O modelo co… | Sobre buscas tradicionais x IAs generativas e uso de textos de livros piratas: "A busca copiava o conteúdo e devolvia leitores. O mode… | ↗Not briefed |
bluesky bluesky-ai | 8d | If you have kids applying to university and they're offered the opportunity to interview on video, or provide a short video independently -… | If you have kids applying to university and they're offered the opportunity to interview on video, or provide a short video independen… | ↗Not briefed |
hn hn-ai | 8d | Inception's Diffusion LLM API Is Shutting Down🔧 OpenRouter | Sad to hear, but I got an email from hello@inceptionlabs.ai announcing the Mercury API that hosts their diffusion models is shutting d… | ↗Not briefed |
hn hn-ai | 8d | Show HN: A better UX for app permission grants on macOS | Hey HN,<p>I maintain a few Mac apps as side projects. Most of them involve permissions of some kind (microphone, screen recording) and… | ↗Not briefed |
bluesky bluesky-ai | 8d | OpenAI Fooled by Russian Pranksters Into Thinking They Were Speaking to Ukrainian Official “We want to have as deep a relationship as we p… | OpenAI Fooled by Russian Pranksters Into Thinking They Were Speaking to Ukrainian Official “We want to have as deep a relationship as… | ↗Not briefed |
hn hn-ai | 8d | Claude the Business Analyst | — | ↗Not briefed |
bluesky bluesky-ai | 8d | same criticisms ive always had which is that what people could & should do is to punch up, see the systemic problem of normalization that y… | same criticisms ive always had which is that what people could & should do is to punch up, see the systemic problem of normalization t… | ↗Not briefed |
bluesky bluesky-ai | 8d | AI companies: They sure do underpin our entire economy. interc.pt/4AkFHSy | AI companies: They sure do underpin our entire economy. interc.pt/4AkFHSy | ↗Not briefed |
hn hn-ai | 8d | The World Builder the AI Labs Haven't Called Yet | — | ↗Not briefed |
bluesky bluesky-ai | 8d | It beggars belief that the American Association of Literary Agents conference was sponsored by an LLM company which received an endorsement… | It beggars belief that the American Association of Literary Agents conference was sponsored by an LLM company which received an endors… | ↗Not briefed |
bluesky bluesky-ai | 8d | I eventually got tired of hitting walls and ditched the tool entirely. I rewrote my test game from scratch and then binned it and started o… | I eventually got tired of hitting walls and ditched the tool entirely. I rewrote my test game from scratch and then binned it and star… | ↗Not briefed |
hn hn-ai | 8d | Show HN: JVM Library to create styled PDFs with easy DSL | — | ↗Not briefed |