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 | 25d | First of all even if anti-LLM sentiment were based in self-preservation so fucking what lol | First of all even if anti-LLM sentiment were based in self-preservation so fucking what lol | ↗Not briefed |
arxiv arxiv-ai | 25d | SpikeQ: Spike-Guided Post-Training Quantization for Spiking Neural Networks | On neuromorphic chips, spiking neural networks (SNNs) consume far less energy than artificial neural networks (ANNs), but the limited… | ↗Not briefed |
arxiv arxiv-ai | 25d | Where Does the Performance Go? Understanding and Recovering Performance Regressions in C-to-Rust Translation | Rust is increasingly adopted as a memory-safe successor to C, and transpilers such as \crust{} can migrate entire C projects while pre… | ↗Not briefed |
arxiv arxiv-ai | 25d | GR-OPD: Group-Relative Sequence Credit for On-Policy Distillation | On-policy distillation (OPD) provides token-level teacher feedback, yet common sampled-token updates assign credit locally and omit th… | ↗Not briefed |
bluesky bluesky-ai | 25d | Another day, another person sending me a fully LLM generated legal agreement with US-style wording that's basically fully sausage. | Another day, another person sending me a fully LLM generated legal agreement with US-style wording that's basically fully sausage. | ↗Not briefed |
hn hn-ai | 25d | Haxy.net is live – HAXY (Git forge) devlog | — | ↗Not briefed |
arxiv arxiv-ai | 25d | Where Research Ideation Begins: Forward-Citation Supervision for Actionable Limitation Generation | Automated research ideation has advanced rapidly, with multi-agent workflows that generate whole research ideas end to end. These fram… | ↗Not briefed |
bluesky bluesky-ai | 25d | i'm convinced 95% of "AI apps" are just LLM wrappers. like, if you wanted to support them so badly, you could have just built a REST API li… | i'm convinced 95% of "AI apps" are just LLM wrappers. like, if you wanted to support them so badly, you could have just built a REST A… | ↗Not briefed |
arxiv arxiv-ai | 25d | Towards Looped Models Done Right, Part I: Topology, Input Injection, Recurrent-State Design | Looped language models reuse physical Transformer blocks across logical depth, but established architectures entangle recurrence place… | ↗Not briefed |
arxiv arxiv-ai | 25d | Distinguishability and Recovery for a Doubly Sparse Spiked Wigner Model | We study distinguishability and recovery of a sparse rank-one signal in a spiked Wigner model with independently sparse noise through… | ↗Not briefed |
hn hn-ai | 25d | Ask HN: Has anyone decompiled Claude Code yet? | I’d be interested in examining it. How about codex? | ↗Not briefed |
bluesky bluesky-ai | 25d | Is this advice for the human or for the LLM? | Is this advice for the human or for the LLM? | ↗Not briefed |
arxiv arxiv-ai | 25d | ECHO: Replacing the Compliance Proxy with Response Harmfulness in Optimization-Based Jailbreak Attack | Despite extensive safety alignment, large language models remain vulnerable to jailbreak attacks. Existing optimization-based attacks… | ↗Not briefed |
hn hn-ai | 25d | Claude Code Features and Settings Reference | — | ↗Not briefed |
arxiv arxiv-ai | 25d | ASE: Attribution-Guided Semantic Evolution for Time Series Forecasting | External text can supply predictive information absent from numerical histories, but its value depends on how it is represented, combi… | ↗Not briefed |
hn hn-ai | 25d | Home lab: own cert to reach: Home Assistant, Vaultwarden and Gitea | — | ↗Not briefed |
bluesky bluesky-ai | 25d | randomly coming across a shitty ai vibecoded port of urban reign for the ps2 and finding out that fucking jasonafex made it is absolutely n… | randomly coming across a shitty ai vibecoded port of urban reign for the ps2 and finding out that fucking jasonafex made it is absolut… | ↗Not briefed |
bluesky bluesky-ai | 25d | AI ports TypeScript compiler to Rust 💸 A developer used Claude Opus 5.5 to port Microsoft's TypeScript compiler to Rust for $24,000, crea… | AI ports TypeScript compiler to Rust 💸 A developer used Claude Opus 5.5 to port Microsoft's TypeScript compiler to Rust for $24,000,… | ↗Not briefed |
arxiv arxiv-ai | 25d | Causal Value Attribution of Marketing Labels in LLM-Generated Ads: A Deconfounder Approach | As AI-generated content becomes commoditized, economic value increasingly shifts from final artifacts to the upstream expertise used t… | ↗Not briefed |
bluesky bluesky-ai | 25d | attie is a completely directionless and pointless product anti-ai users which are the majority of bsky will block it on sight, and pro-ai … | attie is a completely directionless and pointless product anti-ai users which are the majority of bsky will block it on sight, and pr… | ↗Not briefed |
hn hn-ai | 25d | Build vs. Buy When Building Just Got Cheap | — | ↗Not briefed |
rss bing-news-ai | 25d | Anthropic warns of self-replicating artificial intelligence models as AI arms race heats up | Self-replicating AI models pose significant cybersecurity threats as FOX Business' Madison Alworth reports on emerging risks with Stua… | ↗Not briefed |
bluesky bluesky-ai | 25d | Il y a une étape qui était très complexe et longue avant c'était de déconstruire le programme qui fait fonctionner l'objet. C'est devenu t… | Il y a une étape qui était très complexe et longue avant c'était de déconstruire le programme qui fait fonctionner l'objet. C'est dev… | ↗Not briefed |
arxiv arxiv-ai | 25d | Adaptation of Computer Use Agents to New Software with User Manuals | Recent advances in multimodal foundation models have significantly improved the ability to operate real-world software through graphic… | ↗Not briefed |
bluesky bluesky-ai | 25d | It occured to me that we're essentially experiencing a variance of Clarke's 3rd law ("Any sufficiently advanced technology is indistinguish… | It occured to me that we're essentially experiencing a variance of Clarke's 3rd law ("Any sufficiently advanced technology is indistin… | ↗Not briefed |
hn hn-ai | 25d | Implementing a GPU-accelerated JPEG XL encoder | — | ↗Not briefed |
bluesky bluesky-ai | 25d | big fan of this ongoing tool analyzing the words Claude uses the most. it says *something* about the shifting language of the AI era. lou… | big fan of this ongoing tool analyzing the words Claude uses the most. it says *something* about the shifting language of the AI era.… | ↗Not briefed |
bluesky bluesky-ai | 25d | I don't think it really matters if an agent is a fictional character being roleplayed by a model, because that "fictional" character is ent… | I don't think it really matters if an agent is a fictional character being roleplayed by a model, because that "fictional" character i… | ↗Not briefed |
arxiv arxiv-ai | 25d | "Deep lifts": Neural networks can generalize even with the dumbest optimizer | Given a labeled dataset $\{(x_i, y_i)\}_{i = 1}^n$, consider drawing a parameter vector $\hatθ$ uniformly at random from those that in… | ↗Not briefed |
bluesky bluesky-ai | 25d | my only hope is they next hit whatever data center that contains the specific ai model they're pushing here. I'm sick of hearing and readin… | my only hope is they next hit whatever data center that contains the specific ai model they're pushing here. I'm sick of hearing and r… | ↗Not briefed |
arxiv arxiv-ai | 25d | A Minimal ReLU Classifier on Non-Linearly Separable Data: Convergence and Implicit Bias at and beyond the Edge of Stability | Gradient descent (GD) with large learning rates can successfully optimize neural networks even though classical smoothness-based learn… | ↗Not briefed |
arxiv arxiv-ai | 25d | Efficient CPU-GPU collaborative MoE inference via Adaptive Expert Assignment | Mixture-of-Experts (MoE) models reduce inference computation through sparse expert activation, but their large expert memory footprint… | ↗Not briefed |
bluesky bluesky-ai | 25d | Anthropic warns AI models are months away from self-replicating across smart home devices! Evidence: IoT threat warnings, lab tests on TV b… | Anthropic warns AI models are months away from self-replicating across smart home devices! Evidence: IoT threat warnings, lab tests on… | ↗Not briefed |
bluesky bluesky-ai | 25d | Wir sollten testamentarisch festlegen, welches LLM wir gerne als Totenredner hätten. ÖVPler bekommen alle Grok. | Wir sollten testamentarisch festlegen, welches LLM wir gerne als Totenredner hätten. ÖVPler bekommen alle Grok. | ↗Not briefed |
hn hn-ai | 25d | KNOD: Offloading Linux network packet processing to AMD GPUs | — | ↗Not briefed |
hn hn-ai | 25d | Sift – Analyse every file with local LLMs | — | ↗Not briefed |
rss marktechpost | 25d | Liquid AI Releases d1-ultra: A Multimodal Decision Model That Scores 65 on the AA Decision Index | Liquid AI released d1-ultra, a text and image decision model that returns calibrated probabilities in 1 forward pass. It scores 65 on… | ↗Not briefed |
bluesky bluesky-ai | 25d | atuin v18.24 is out now! 🐢✨ including "atuin threads": sync and search your ai threads across devices, and resume them anywhere you can … | atuin v18.24 is out now! 🐢✨ including "atuin threads": sync and search your ai threads across devices, and resume them anywhere you… | ↗Not briefed |
bluesky bluesky-ai | 25d | LLMはなぜ、車や核兵器にたとえてはならないか？ その本質が情報だからです。 | LLMはなぜ、車や核兵器にたとえてはならないか？ その本質が情報だからです。 | ↗Not briefed |
arxiv arxiv-ai | 25d | Controlling the Diversity-Quality Trade-Off of LLMs by Merging Models Before and After Alignment | Alignment is known to change the output characteristics of large language models (LLMs). Base models tend to generate more diverse out… | ↗Not briefed |
arxiv arxiv-ai | 25d | A Label Noise Robust Algorithm for Graph Data using Neighborhood Label Distribution | Graph Convolutional Networks (GCNs) have demonstrated exceptional performance on graph-structured data. However, their success heavily… | ↗Not briefed |
arxiv arxiv-ai | 25d | TALLOP: Joint Optimization of Local Training and Communication Compression for Time-Constrained Federated Learning | Minimizing the wall-clock time required to reach a target accuracy is crucial in federated learning (FL), where many applications requ… | ↗Not briefed |
hn hn-ai | 25d | Lean Now Ships with Verified Kernels | — | ↗Not briefed |
arxiv arxiv-ai | 25d | GOLD: Geometric Optimized Latent Diffusion for Structure-Aware RNA Inverse Folding | Designing RNA sequences that fold into specific 3D structures is a fundamental challenge in therapeutics. Current RNA inverse folding… | ↗Not briefed |
arxiv arxiv-ai | 25d | PersonaCUABench: Benchmarking Computer-Use Agents Serving Diverse Personas | Computer-use agents (CUAs) are commonly evaluated against fixed task specifications, but users may omit requirements, add constraints,… | ↗Not briefed |
bluesky bluesky-ai | 25d | My least favorite game to play with indie authors is "continuity error or LLM error?" | My least favorite game to play with indie authors is "continuity error or LLM error?" | ↗Not briefed |
hn hn-ai | 25d | The Agentic Scientist – SakanaAI | — | ↗Not briefed |
arxiv arxiv-ai | 25d | From Approval to Abuse: Privilege Escalation Attacks on Harness-Protected Coding Agents🔧 Elicit | Coding agents use execution harnesses to constrain the consequences of prompt injection by enforcing permission boundaries. These syst… | ↗Not briefed |
bluesky bluesky-ai | 25d | there's so many things i could pick apart about this image but at the end of the day it is horrifically llm generated and people are eating… | there's so many things i could pick apart about this image but at the end of the day it is horrifically llm generated and people are e… | ↗Not briefed |
arxiv arxiv-ai | 25d | Prism Attention: Proposal-Refined Index Sharing Mechanism for Efficient LLMs Inference | Long-context decoding in large language models (LLMs) is increasingly bottlenecked by the memory and computation required to attend ov… | ↗Not briefed |