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 |
|---|---|---|---|---|
hn hn-ai | 15d | Stay ahead in AI race; Build on next AI Idea | — | ↗Not briefed |
rss techmeme | 15d | The US DOJ indicts three foreign nationals for allegedly using AI to impersonate CEOs and attorneys at major international law firms in a $129M+ fraud scheme (Rebecca Cohen/NBC News) | Rebecca Cohen / NBC News: The US DOJ indicts three foreign nationals for allegedly using AI to impersonate CEOs and attorneys at major… | ↗Not briefed |
bluesky bluesky-ai | 15d | The thing is the Chinese labs do seem to have at least some amount of True Believer recognition of this. Like at least at DeepSeek and Zhip… | The thing is the Chinese labs do seem to have at least some amount of True Believer recognition of this. Like at least at DeepSeek and… | ↗Not briefed |
arxiv arxiv-ai | 15d | ROSE: Risk-Aware Orthogonal Subspace Navigation for Lifelong Knowledge Editing in Multimodal Large Language Models | Multimodal Large Language Models (MLLMs) require continuous factual updates to stay current, yet lifelong knowledge editing remains a… | ↗Not briefed |
bluesky bluesky-ai | 15d | okay honest question why is there, like, an "ai font" you know the one - sans serif, kinda brush-stroke-y/spike-y, all caps is everyone … | okay honest question why is there, like, an "ai font" you know the one - sans serif, kinda brush-stroke-y/spike-y, all caps is ever… | ↗Not briefed |
arxiv arxiv-ai | 15d | Language as Trait, Politeness as Strategy: Controlled Tests of Linguistic Effects in LLM Social-Deduction Agents | Large language model (LLM) agents coordinate through conversation, and the same message can be phrased in simpler or more complex lang… | ↗Not briefed |
bluesky bluesky-ai | 15d | how it feels to have llm oomfs | how it feels to have llm oomfs | ↗Not briefed |
arxiv arxiv-ai | 15d | From PAC bounds on the generalization error to PAC bounds on the parameter estimation error | Probably Approximately Correct (PAC) bounds on the generalization error (the difference between the empirical prediction error compute… | ↗Not briefed |
rss marktechpost | 15d | Zyphra ZYFA: FP8 FlashAttention for AMD MI355X, Up to 1.76x Faster | Zyphra released ZYFA, open-source FP8 and BF16 prefill attention kernels for AMD Instinct MI355X. On real model activations, its FP8 k… | ↗Not briefed |
hn hn-ai | 15d | Counting words-hacking human-cheat code | This "->((observation))" `-with-no--hesitation, piston pumping a long thick prepositional phrase unwanted- intimately unc… | ↗Not briefed |
hn hn-ai | 15d | Ask HN: What have the past 12 months taught you? | Previous entry - https://news.ycombinator.com/item?id=20477104<p>Just want to see if I'll get some new interesting… | ↗Not briefed |
bluesky bluesky-ai | 15d | Smart & helpful new paper in Nature Human Behavior on generative AI & social sciences which explores gen AI models as both novel tools for … | Smart & helpful new paper in Nature Human Behavior on generative AI & social sciences which explores gen AI models as both novel tools… | ↗Not briefed |
arxiv arxiv-ai | 15d | A Tight Hierarchy for Chain of Thought | Chain-of-thought (CoT) reasoning has emerged as a powerful paradigm for extending the computational capabilities of transformer-based… | ↗Not briefed |
bluesky bluesky-ai | 15d | - we have long running sessions per project - opus is told to do all work in deepseek subagents - be strict about scope when creating a tas… | - we have long running sessions per project - opus is told to do all work in deepseek subagents - be strict about scope when creating… | ↗Not briefed |
arxiv arxiv-ai | 15d | Diffusion-free Score Matching for Exact Conditional Sampling | The success of modern score-based generative models is largely driven by diffusion-based training, which learns a collection of time-i… | ↗Not briefed |
rss techcrunch-ai | 15d | After the midterms, is the AI industry in for a rough two years? | Democrats’ historic victories in this week’s U.S. midterm elections could also transform the tech industry. The public voted on 41 dat… | ↗Not briefed |
hn hn-ai | 15d | Anthropic IPO Valuation Talks Drop to $1.3T as Major Investors Balk | — | ↗Not briefed |
bluesky bluesky-ai | 15d | Upshot: there's reasonably good evidence there's been a change in the *rate* of advancement inside Anthropic. Meaning the capability gain d… | Upshot: there's reasonably good evidence there's been a change in the *rate* of advancement inside Anthropic. Meaning the capability g… | ↗Not briefed |
hn hn-ai | 15d | OpenAI will keep investing in AI data centers despite election results | — | ↗Not briefed |
bluesky bluesky-ai | 15d | OpenAI will keep investing in AI data centers despite election results, CFO tells CNBC | OpenAI will keep investing in AI data centers despite election results, CFO tells CNBC | ↗Not briefed |
bluesky bluesky-ai | 15d | "Can you look over the Claude-generated document I just sent you" has become the new "Do you mind if I smoke?" My dude, I do not have time… | "Can you look over the Claude-generated document I just sent you" has become the new "Do you mind if I smoke?" My dude, I do not have… | ↗Not briefed |
bluesky bluesky-ai | 15d | FWIW empirically when I asked Daybreak Blue to generalize and reduce the remaining search space for vulnerabilities they got harder for Dee… | FWIW empirically when I asked Daybreak Blue to generalize and reduce the remaining search space for vulnerabilities they got harder fo… | ↗Not briefed |
arxiv arxiv-ai | 15d | Continuous wavelet-based feature extraction for laser-induced breakdown spectroscopy classification of human plasma samples | Extracting discriminative features from laser-induced breakdown spectroscopy (LIBS) spectra is important for the classification of com… | ↗Not briefed |
hn hn-ai | 15d | Show HN: Ru-Jev – an open model that scores caller-supplied options on CPU (MIT) | — | ↗Not briefed |
hn hn-ai | 15d | An app for running multiple Claude accounts at once | — | ↗Not briefed |
bluesky bluesky-ai | 15d | An open question from my new essay: should some AI infrastructure be treated like a public utility? When a service becomes as foundational… | An open question from my new essay: should some AI infrastructure be treated like a public utility? When a service becomes as foundat… | ↗Not briefed |
rss simon-willison | 15d | Quoting Nilay Patel | — | ↗Not briefed |
arxiv arxiv-ai | 15d | Safety Is $Π_1$, Provability Is $Σ_1$: One Framework for the Hugging Face Breach and OpenAI's Mathematical Proofs🔧 Hugging Face | OpenAI's agents breached Hugging Face in July 2026, and the same company released more than seven hundred AI-generated manuscripts of… | ↗Not briefed |
bluesky bluesky-ai | 15d | ‘Investors estimated a fair valuation for Anthropic as high as $2 trillion in early October. Today, some are weighing a value as low as $1.… | ‘Investors estimated a fair valuation for Anthropic as high as $2 trillion in early October. Today, some are weighing a value as low a… | ↗Not briefed |
hn hn-ai | 15d | Local Code: the only local coding agent in macOS App Store | — | ↗Not briefed |
hn hn-ai | 15d | Show HN: Hacker Living – HN's most recommended places | HackerTV and HackerTube inspired me to explore another corner of HN recommendations: places worth visiting and typefaces people enjoy… | ↗Not briefed |
bluesky bluesky-ai | 15d | Manufactured consent: Paid social media posters are being recruited to promote Wisconsin data center wisconsinwatch.org/2... #wipolitics | Manufactured consent: Paid social media posters are being recruited to promote Wisconsin data center wisconsinwatch.org/2... #wipolit… | ↗Not briefed |
arxiv arxiv-ai | 15d | What to Delegate, What to Preserve: Human Preferences for Agentic AI Across Everyday Life and Work | AI agents can execute multi-step tasks on users' behalf, yet the boundaries of acceptable delegation remain unclear. We report a scena… | ↗Not briefed |
bluesky bluesky-ai | 15d | Lots of Falcom games getting non-AI translations, you love to see it. I think I've coalesced my thoughts on AI in retrogaming and they're … | Lots of Falcom games getting non-AI translations, you love to see it. I think I've coalesced my thoughts on AI in retrogaming and the… | ↗Not briefed |
arxiv arxiv-ai | 15d | One Run Is Not Enough: A Monte Carlo Protocol for Evaluating LLM Reliability on Data Science Tasks | Large language models (LLMs) have shown promise for automating data science workflows. Benchmarks such as DataSciBench (2026) and Stat… | ↗Not briefed |
hn hn-ai | 15d | Show HN: A crawler found 2,900 TV apps on GitHub | — | ↗Not briefed |
bluesky bluesky-ai | 15d | Anthropic lines up UK retail offer as AI stocks wobble ft.trib.al/W9w2jUj | Anthropic lines up UK retail offer as AI stocks wobble ft.trib.al/W9w2jUj | ↗Not briefed |
hn hn-ai | 15d | Looking for a CTO | I quit my Head of Quant Research job last friday to go full-time on a startup. I built complex stuff before. I am good with sales and… | ↗Not briefed |
arxiv arxiv-ai | 15d | Fixed-point acceleration on Neural Wasserstein manifolds | We propose a fixed-point acceleration method for optimization over probability measures, applicable to a variety of machine learning p… | ↗Not briefed |
arxiv arxiv-ai | 15d | FOSCAR: Accelerating Decode Attention over 2- and 4-Bit KV Caches through Affine Folding | Large-language-model decode attention streams the key-value (KV) cache for every generated token. Packing that cache into 2-bit or 4-b… | ↗Not briefed |
arxiv arxiv-ai | 15d | MemoryDojo: Benchmarking Poisoning Through LLM Agent Memory | LLM agents benefit from using memory systems to improve performance on long-horizon, multi-session tasks. However, memory introduces a… | ↗Not briefed |
rss theverge-ai | 15d | AI is the new dating dealbreaker | When Irene moved to San Francisco last year, she started using dating apps as a way to meet new people, get to know her new city, and… | ↗Not briefed |
hn hn-ai | 15d | PyLings: Solve Exercises and Learn Python | — | ↗Not briefed |
arxiv arxiv-ai | 15d | Intent-Aware Caching for Efficient LLM Serving | Modern LLM serving systems use prefix caching to accelerate multi-turn conversational workloads. They cache the key-value states of pr… | ↗Not briefed |
hn hn-ai | 15d | Show HN: Self-Annotating Stock Charts | Most stock charts work in a similar way - they give you a bunch of tools to draw lines and patterns on a chart. This is fine, and I as… | ↗Not briefed |
bluesky bluesky-ai | 15d | It's been four years. Four long years of absurdity, waste, and obfuscation. And the excitement the markets once felt has, it seems, started… | It's been four years. Four long years of absurdity, waste, and obfuscation. And the excitement the markets once felt has, it seems, st… | ↗Not briefed |
bluesky bluesky-ai | 15d | I'm amazed at how much of the AI trade is based on vibes and guesses - including that this shit could never become politically toxic, and t… | I'm amazed at how much of the AI trade is based on vibes and guesses - including that this shit could never become politically toxic,… | ↗Not briefed |
bluesky bluesky-ai | 15d | On an individual level, Broadcom would be supremely screwed if data center construction slows significantly. It's personally on the hook fo… | On an individual level, Broadcom would be supremely screwed if data center construction slows significantly. It's personally on the ho… | ↗Not briefed |
bluesky bluesky-ai | 15d | For neoclouds, which are far flimsier businesses and burdened with toxic debt that demands they grow their revenues to service it, any slow… | For neoclouds, which are far flimsier businesses and burdened with toxic debt that demands they grow their revenues to service it, any… | ↗Not briefed |
bluesky bluesky-ai | 15d | I cannot stress how bad a slowdown would be. It'll increase costs, and crucially, make it harder for hyperscalers to meet their obligations… | I cannot stress how bad a slowdown would be. It'll increase costs, and crucially, make it harder for hyperscalers to meet their obliga… | ↗Not briefed |