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 | 4d | The Geometry of Meaning in Neural Manifolds🔧 Modal | Just as different languages express the same concept through different words, neural networks express similar functions through incomp… | ↗Not briefed |
arxiv arxiv-ai | 4d | Bridging Number-Theoretic Primality Testing And Algorithmic Optimization: An Adaptive Cryptographic Framework | The security of modern public-key cryptographic systems, such as RSA, Diffie-Hellman, and ECC, depends entirely on the computational i… | ↗Not briefed |
hn hn-ai | 4d | Show HN: NeoInbox, a recreation of Google Inbox as a Thunderbird extension | 12 years ago today (Oct 22, 2014), Google released Inbox by Gmail™, definitively solving email UX. Even now, 7 years after Google made… | ↗Not briefed |
arxiv arxiv-ai | 4d | Verbalizing Activation Differences with Natural Language Autoencoders | Activation verbalizers translate the internal representations of large language models into human-readable text. The recently proposed… | ↗Not briefed |
arxiv arxiv-ai | 4d | ReflectPO: Reflection-Guided Policy Optimization for Long-Horizon LLM Agents | With external tools, large language models (LLMs) can tackle complex tasks through sustained environmental interaction. LLM-based agen… | ↗Not briefed |
hn hn-ai | 4d | Achieving AI-style reasoning without an LLM | — | ↗Not briefed |
bluesky bluesky-ai | 4d | A special fuck you to Chainguard for breaking the embargo and being the reason we're seeing active exploits in the wild before arm64 builds… | A special fuck you to Chainguard for breaking the embargo and being the reason we're seeing active exploits in the wild before arm64 b… | ↗Not briefed |
bluesky bluesky-ai | 4d | Sometimes I feel like I talk like an LLM when autism masking or trying to damage control tension building in a social situation (something … | Sometimes I feel like I talk like an LLM when autism masking or trying to damage control tension building in a social situation (somet… | ↗Not briefed |
bluesky bluesky-ai | 4d | Pandoras box has been opened. Even if anthropic and openai shut down tomorrow, Chinese labs could continue putting out open source models. …🔧 Replicate | Pandoras box has been opened. Even if anthropic and openai shut down tomorrow, Chinese labs could continue putting out open source mod… | ↗Not briefed |
bluesky bluesky-ai | 4d | In cyberpunk lore, Bartmoss released RABIDS causing the datakrash in 2022 which caused the death of the net and everything to devolve into … | In cyberpunk lore, Bartmoss released RABIDS causing the datakrash in 2022 which caused the death of the net and everything to devolve… | ↗Not briefed |
bluesky bluesky-ai | 4d | RT カプコンが標的になったか 正直な意見を言わせて貰えば心底どうでもいいよ 生成AI/LLMに親和的態度の企業には愛想を尽かしているからどうなっても知ったことではない | RT カプコンが標的になったか 正直な意見を言わせて貰えば心底どうでもいいよ 生成AI/LLMに親和的態度の企業には愛想を尽かしているからどうなっても知ったことではない | ↗Not briefed |
arxiv arxiv-ai | 4d | Low-Latency MoE Inference in Space Data Centers: A Wireless-Power-Transfer Perspective | Space data centers (SpaceDCs) are emerging as platforms for large language models (LLMs), supported by continuous solar-energy harvest… | ↗Not briefed |
rss mit-tech-review-ai | 4d | AI’s hidden storage bill: What inference leaves behind | Ask most executives to picture the cost of AI and they’ll likely see GPUs—racks of accelerators with gigawatts of power—and a waitlist… | ↗Not briefed |
arxiv arxiv-ai | 4d | DConstrain: Compiling Context-Parameterized Transitions for Diffusion Language Model Constrained Decoding | Diffusion language models (DLMs) decode multiple tokens in parallel, so individually valid tokens can be incompatible together under a… | ↗Not briefed |
arxiv arxiv-ai | 4d | Harnessing Language Models as Optimization Modeling Agents | Large language models (LLMs) can translate natural-language optimization problems into solver programs, but reliably producing correct… | ↗Not briefed |
rss the-decoder | 4d | Anthropic takes on Grok Bot, Meta's Muse, and OpenAI's Dots with its always-on agent "Claude Duet" | Anthropic has launched Claude Duet, a proactive AI agent that handles tasks on its own, even when the user is offline. The app is buil… | ↗Not briefed |
arxiv arxiv-ai | 4d | MIG-COW: Transferable Adversarial Attacks on Deepfake Detectors via Gradient Decomposition🔧 Consensus | Despite recent advances, deepfake detectors remain vulnerable to adversarial examples, particularly in diverse, real-world settings. W… | ↗Not briefed |
arxiv arxiv-ai | 4d | Hierarchical Fine-Grained Cross-Modal Alignment via Dynamic Granularity Discovery and Optimal Transport🔧 Modal | Fine-grained image-text retrieval requires a model not only to assess global semantic consistency between an image and a text, but als… | ↗Not briefed |
bluesky bluesky-ai | 4d | Was randomly listening to a history podcast. They have an Anthropic ad in the middle of the episode and call Columbus a pioneer. I check wh… | Was randomly listening to a history podcast. They have an Anthropic ad in the middle of the episode and call Columbus a pioneer. I che… | ↗Not briefed |
arxiv arxiv-ai | 4d | TwinStore: A Synthetic Consumer Platform Using Vision--Language Models | We present TwinStore, a source-available synthetic consumer platform for physical retail built on vision--language models. TwinStore t… | ↗Not briefed |
arxiv arxiv-ai | 4d | Query-Time Calibration of CLIP Features for Open-Vocabulary 3D Segmentation | Lifting vision-language features such as CLIP into 3D Gaussian reconstructions enables open-vocabulary segmentation through natural-la… | ↗Not briefed |
bluesky bluesky-ai | 4d | Hecho y te pongo literalmente la opinión de mi psiquiatra sobre la IA como "terapeuta": [la IA no confronta y eso es potencialmente peligro… | Hecho y te pongo literalmente la opinión de mi psiquiatra sobre la IA como "terapeuta": [la IA no confronta y eso es potencialmente pe… | ↗Not briefed |
arxiv arxiv-ai | 4d | Same molecule, Different Knowledge: Cross-Identifier Binding Failures in LLMs For chemistry | Chemical AI workflows routinely represent the same molecule through multiple co-referring identifiers, including trivial names, IUPAC… | ↗Not briefed |
arxiv arxiv-ai | 4d | Beyond Phenotypic Similarity: Relation Refinement and Selective Population Graph Learning for Multimodal Brain Disorder Diagnosis | Population graph learning supports brain disorder diagnosis by sharing information across subjects. Its effectiveness, however, depend… | ↗Not briefed |
arxiv arxiv-ai | 4d | Silicon Samples Match the Answers but Not the Respondents | Silicon sampling generates survey respondents by having a large language model (LLM) answer a survey as a person with given demographi… | ↗Not briefed |
arxiv arxiv-ai | 4d | Parameter Learning in Active Inference via Variational Message Passing | Active inference casts perception, learning and planning as variational inference, where the Expected Free Energy (EFE) defines the pl… | ↗Not briefed |
bluesky bluesky-ai | 4d | at this point I'd be *happy* about llm-written documentation as long as it exists on a page in a wiki. having non-deterministic answers onl… | at this point I'd be *happy* about llm-written documentation as long as it exists on a page in a wiki. having non-deterministic answer… | ↗Not briefed |
hn hn-ai | 4d | Ask HN: Anyone patching servers by prompting AI instead of SSH/remote desktop? | I did exactly this yesterday and it worked great. Not super critical systems, but almost 10 devices, so it saved me a lot of time. I u… | ↗Not briefed |
bluesky bluesky-ai | 4d | Jep. Tämä esim. tekee tilanteen, jossa LinkedIn kontaktit halusi päältä vuosi sitten auttaa minua, etten joutuisi vakavasti sairaana kadull… | Jep. Tämä esim. tekee tilanteen, jossa LinkedIn kontaktit halusi päältä vuosi sitten auttaa minua, etten joutuisi vakavasti sairaana k… | ↗Not briefed |
hn hn-ai | 4d | Show HN: BlueFacts – Bluesky stats and feeds via MCP | I work on BlueFacts a stats site for Bluesky, it tracks accounts, custom feeds, tags etc<p>I added MCP support yesterday so you can cr… | ↗Not briefed |
bluesky bluesky-ai | 4d | This episode of Whyyyyy is brought to you by the letter PHP and the number 4 Wishing every critter out there that has to deal with this go… | This episode of Whyyyyy is brought to you by the letter PHP and the number 4 Wishing every critter out there that has to deal with th… | ↗Not briefed |
bluesky bluesky-ai | 4d | AI copyright lawsuits are coming thick and fast. In just the last couple of weeks: - USA Today & 18 other papers sued OpenAI - around 300 … | AI copyright lawsuits are coming thick and fast. In just the last couple of weeks: - USA Today & 18 other papers sued OpenAI - around… | ↗Not briefed |
bluesky bluesky-ai | 4d | PortedGames: PortedGames is a directory of fan-made and AI-assisted ports of classic games: browser ports you can play online, and native P… | PortedGames: PortedGames is a directory of fan-made and AI-assisted ports of classic games: browser ports you can play online, and nat… | ↗Not briefed |
bluesky bluesky-ai | 4d | 数千体のAIが数百万個の脳細胞データをわずか32分で分析、アルツハイマー病やパーキンソン病の研究につながる新たな手掛かりを発見 https://gigazine.net/news/20261022-openai-research-complex-biological-data/ | 数千体のAIが数百万個の脳細胞データをわずか32分で分析、アルツハイマー病やパーキンソン病の研究につながる新たな手掛かりを発見 https://gigazine.net/news/20261022-openai-research-complex-biological-… | ↗Not briefed |
arxiv arxiv-ai | 4d | Equilibria and Learning in Competitive Market Making | We propose a market-making game played among $N \ge 2$ strategic market makers and a single non-strategic representative trader. In a… | ↗Not briefed |
arxiv arxiv-ai | 4d | Continuous-depth Deep Gaussian Processes | Deep Gaussian processes (DGPs) are appealing Bayesian regression models, but their standard discrete-depth parameterization often beco… | ↗Not briefed |
arxiv arxiv-ai | 4d | Building Presentation Intelligence for High-Quality Slide Generation | PowerPoint presentations (PPTs) are a dominant medium for communication across diverse domains, yet high-quality PPT generation still… | ↗Not briefed |
arxiv arxiv-ai | 4d | From Perpetrators' Digital Footprints to Behavioral Communities: An Ontology-Driven Framework for Bengali Cyberbullying Analysis--- FULL VERSION | Cyberbullying involves behavioral and contextual patterns beyond isolated abusive messages. However, existing approaches remain fragme… | ↗Not briefed |
arxiv arxiv-ai | 4d | Tracing Subliminal Learning Through Pretraining | Large language models trained on data generated by other models can sometimes inherit traits unrelated to that data, a phenomenon call… | ↗Not briefed |
bluesky bluesky-ai | 4d | @laurencerouil.bsky.social, #CAMS Director, introduces AIFS-COMPO, an experimental system forecasting ozone, particulate matter, aerosol op… | @laurencerouil.bsky.social, #CAMS Director, introduces AIFS-COMPO, an experimental system forecasting ozone, particulate matter, aeros… | ↗Not briefed |
arxiv arxiv-ai | 4d | Inherited, Not Invented: Tracing the Provenance of Wait in Reasoning Models | Reinforcement learning (RL) is credited with teaching reasoning models to write {wait} before reconsidering a step, yet large RL accur… | ↗Not briefed |
bluesky bluesky-ai | 4d | Copilot, generate me an ad that sells our LLM services to Patrick Bateman from American Psycho | Copilot, generate me an ad that sells our LLM services to Patrick Bateman from American Psycho | ↗Not briefed |
rss techmeme | 4d | iPhone Duo review: the outer device is by far the worst iPhone Apple has ever made, while the inner device is a splendid tablet with a UI that bests the iPad's (John Gruber/Daring Fireball) | John Gruber / Daring Fireball: iPhone Duo review: the outer device is by far the worst iPhone Apple has ever made, while the inner dev… | ↗Not briefed |
arxiv arxiv-ai | 4d | One Teacher, Many Views: Reducing Reference Sensitivity in On-Policy Self-Distillation | On-policy self-distillation (OPSD) improves reasoning by training a model on its own trajectories with dense token-level supervision f… | ↗Not briefed |
bluesky bluesky-ai | 4d | Had so much fun at #JDD 2026 in Kraków yesterday. I presented “Java Goes CUDA-Native”: new @tornadovm.org features that let @jitllm.io run… | Had so much fun at #JDD 2026 in Kraków yesterday. I presented “Java Goes CUDA-Native”: new @tornadovm.org features that let @jitllm.i… | ↗Not briefed |
bluesky bluesky-ai | 4d | Neat, costly patterns; are not reality. Truth is not stochastic. #Senryū #Poem #Poetry #μvers #micropoetry #LLM #Vibe #Forecasting #Fast … | Neat, costly patterns; are not reality. Truth is not stochastic. #Senryū #Poem #Poetry #μvers #micropoetry #LLM #Vibe #Forecasting #… | ↗Not briefed |
arxiv arxiv-ai | 4d | MOER: Mix-Policy Optimization under Expert Prefix Guidance with Selective Rollback | Mix-policy optimization incorporates off-policy expert data into the reinforcement learning (RL) loop to transcend the capability boun… | ↗Not briefed |
bluesky bluesky-ai | 4d | まあその質問した人に価値があるないじゃなくて本当はそんなくだらないことに大量の資源と権利を吐き捨てるLLMに価値がないんですけどね | まあその質問した人に価値があるないじゃなくて本当はそんなくだらないことに大量の資源と権利を吐き捨てるLLMに価値がないんですけどね | ↗Not briefed |
bluesky bluesky-ai | 4d | 趣味探しすらLLMに検索させる人間に一体なんの価値が？？？？ | 趣味探しすらLLMに検索させる人間に一体なんの価値が？？？？ | ↗Not briefed |
rss huggingface-blog | 4d | Ensemble forecasting with IBM Granite time-series models | — | ↗Not briefed |