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 | 5d | Governance-State Corruption in Long-Term Memory for LLM Agents: A Cross-Model Stress Test | An LLM agent can retrieve a memory correctly yet use it incorrectly after its validity state changes. We study governance-state corrup… | ↗Not briefed |
bluesky bluesky-ai | 5d | the pro ai accounts on here don't work for anthropic and they do a better job at pr than the people who are paid by them to talk about it o… | the pro ai accounts on here don't work for anthropic and they do a better job at pr than the people who are paid by them to talk about… | ↗Not briefed |
bluesky bluesky-ai | 5d | opencode has deepseek flash for free and it’ll happily redteam your infra without issue. you’re irresponsible if you don’t try it. | opencode has deepseek flash for free and it’ll happily redteam your infra without issue. you’re irresponsible if you don’t try it. | ↗Not briefed |
arxiv arxiv-ai | 5d | Disentangling Robustness and Sycophancy in Multilingual LLMs | When a multilingual LLM agrees with a user's false premise, is it being sycophantic, or does it simply not know better in that languag… | ↗Not briefed |
hn hn-ai | 5d | Open Language Models – Nathan Lambert | — | ↗Not briefed |
bluesky bluesky-ai | 5d | "people who are mad at the Internet Archive for using Claude to fix their website because LLM training is 'plagiarism'" is an absolutely IN… | "people who are mad at the Internet Archive for using Claude to fix their website because LLM training is 'plagiarism'" is an absolute… | ↗Not briefed |
bluesky bluesky-ai | 5d | okay the swift rewrite of the bluesky app is VERY good it cost about 40 dollars in deepseek tokens then about 150-200 in sonnet tokens to … | okay the swift rewrite of the bluesky app is VERY good it cost about 40 dollars in deepseek tokens then about 150-200 in sonnet token… | ↗Not briefed |
arxiv arxiv-ai | 5d | LoRA Superposition across Pretrained Models | Low-rank adaptation (LoRA) is usually tied to the pretrained backbone on which it is learned. We ask whether LoRAs trained independent… | ↗Not briefed |
arxiv arxiv-ai | 5d | From Clinical Judgment to Checkable Rules: Clinician-Led Guardrails for Mental-Health Chatbots | Large language models (LLMs) deployed in mental-health contexts can be evaluated for safety failures, but the resulting judgments are… | ↗Not briefed |
bluesky bluesky-ai | 5d | 文化についてAIに質問すると日本が選ばれやすい傾向がある、8モデル・24言語で回答の偏りを調査 - GIGAZINE 2026年10月22日 gigazine.net/news/2026102... "たとえば、中国はデータに最も多く登場していたモデルでも回答にはほとんど現… | 文化についてAIに質問すると日本が選ばれやすい傾向がある、8モデル・24言語で回答の偏りを調査 - GIGAZINE 2026年10月22日 gigazine.net/news/2026102... "たとえば、中国はデータに最も多く登場していたモデルでも回答には… | ↗Not briefed |
hn hn-ai | 5d | Advice You'll Get Is the Advice You've Heard | — | ↗Not briefed |
bluesky bluesky-ai | 5d | Anthropic pitched a draft federal AI safety bill to Congress proposing to block models carrying a foreseeable risk of 50+ deaths or $1B+ in… | Anthropic pitched a draft federal AI safety bill to Congress proposing to block models carrying a foreseeable risk of 50+ deaths or $1… | ↗Not briefed |
bluesky bluesky-ai | 5d | OpenAI抛出的几百个数学问题证明还没被数学家证实是成立的（我看到有美国数学家说其证明不知所云，纯属搅局），清华的数学教授却认为那是因为AI审核能力超过数学家，极高的概率会是正确的，是数学家太笨了。没有基本的逻辑思辨能力也能当数学教授，还迷信人工智能是超级智能，中国奇迹。 | OpenAI抛出的几百个数学问题证明还没被数学家证实是成立的（我看到有美国数学家说其证明不知所云，纯属搅局），清华的数学教授却认为那是因为AI审核能力超过数学家，极高的概率会是正确的，是数学家太笨了。没有基本的逻辑思辨能力也能当数学教授，还迷信人工智能是超级智能，中国… | ↗Not briefed |
rss marktechpost | 5d | NVIDIA Nsight Python Tutorial: How to Profile GPU Kernels | Discover NVIDIA Nsight Python, an open-source interface that allows developers to profile GPU kernels, analyze hardware metrics, evalu… | ↗Not briefed |
arxiv arxiv-ai | 5d | An Agent Architecture with Tools and Memory for Selective Correction of Wrist PPG Heart Rate Estimates | Wrist photoplethysmography (PPG) heart rate estimates are vulnerable to motion artifacts, and corrections that reduce mean error can w… | ↗Not briefed |
hn hn-ai | 5d | MiniForth – A small Lean tutorial for programming language semantics | — | ↗Not briefed |
bluesky bluesky-ai | 5d | i should probably not be allowed to point "/goal get remote shell" deepseek 1017 at things this feels like having a nuclear weapon | i should probably not be allowed to point "/goal get remote shell" deepseek 1017 at things this feels like having a nuclear weapon | ↗Not briefed |
arxiv arxiv-ai | 5d | Learning Domain Trajectories with Flow Matching for Gradual Domain Adaptation | Large distribution shifts remain a primary obstacle for domain generalization and unsupervised adaptation. Although gradual domain ada… | ↗Not briefed |
bluesky bluesky-ai | 5d | Fondly remembering that time Bob McCoskrie spent hours of his life arguing with a pre-llm twitter bot that just regurgitated nonsense based… | Fondly remembering that time Bob McCoskrie spent hours of his life arguing with a pre-llm twitter bot that just regurgitated nonsense… | ↗Not briefed |
arxiv arxiv-ai | 5d | Direct cueing of word identity through word duration: evidence from Mandarin homophones | While there is increasing evidence that words are stored in the mental lexicon with some phonetic detail, there are open questions abo… | ↗Not briefed |
rss techmeme | 5d | Walmart unveils its Agent Gateway, a new MCP server to help users' approved AI agents shop its website, and dedicated Agent Profiles with special "terms" for AI (Sarah Perez/TechCrunch) | Sarah Perez / TechCrunch: Walmart unveils its Agent Gateway, a new MCP server to help users' approved AI agents shop its website, and… | ↗Not briefed |
bluesky bluesky-ai | 5d | LLM 이거 코딩말고는 정말 노쓸모인 것 같아. | LLM 이거 코딩말고는 정말 노쓸모인 것 같아. | ↗Not briefed |
bluesky bluesky-ai | 5d | just saw a horrific genAI political ad for some voter prop in california. we need to draw and quarter everyone on the board of anthropic etc | just saw a horrific genAI political ad for some voter prop in california. we need to draw and quarter everyone on the board of anthrop… | ↗Not briefed |
bluesky bluesky-ai | 5d | PHP問題は「一部の馬鹿がリークしたせいでLLM(生成AI)で脆弱箇所突かれそうだから気をつけて慎重にパッチ当てろ」って感じでしたか… | PHP問題は「一部の馬鹿がリークしたせいでLLM(生成AI)で脆弱箇所突かれそうだから気をつけて慎重にパッチ当てろ」って感じでしたか… | ↗Not briefed |
bluesky bluesky-ai | 5d | You know, the thing that fucking gets me about software devs having fully drank the Openai koolaid, is that there's literally no fucking be… | You know, the thing that fucking gets me about software devs having fully drank the Openai koolaid, is that there's literally no fucki… | ↗Not briefed |
bluesky bluesky-ai | 5d | LLMにコードレビューさせるとたぶんめちゃんこまともな指摘するんだろうけど、同じことLLMにスクラッチでHTTPサーバ書かせちゃりするとガバガバの実装吐き出してきそう、みちゃいなイメージがあるかちゃ | LLMにコードレビューさせるとたぶんめちゃんこまともな指摘するんだろうけど、同じことLLMにスクラッチでHTTPサーバ書かせちゃりするとガバガバの実装吐き出してきそう、みちゃいなイメージがあるかちゃ | ↗Not briefed |
arxiv arxiv-ai | 5d | Scaling Laws for Safety Probes | Safety probes trained on a model's activations are a standard tool for monitoring language models in real time. Each time a new model… | ↗Not briefed |
rss bing-news-ai | 5d | Australian families turn to artificial intelligence in fight against soaring supermarket prices | Australian families are turning to artificial intelligence to combat soaring grocery costs, with new apps helping shoppers save hundre… | ↗Not briefed |
rss bing-news-ai | 5d | Australian families turn to artificial intelligence in fight against soaring supermarket prices | Australian families are turning to artificial intelligence to combat soaring grocery costs, with new apps helping shoppers save hundre… | ↗Not briefed |
bluesky bluesky-ai | 5d | ・ここ最近の企業へのサイバー攻撃多発には間違いなくAIが使用されているが、そちらについては「世界はいくつかの悪い出来事が発生することを受け入れるべきだと考えている」とCEOが発言し責任を取る姿勢を全く示さなかったのに対し、こういう時ばかり「今回の脆弱性を発見したのは自社のAIで… | ・ここ最近の企業へのサイバー攻撃多発には間違いなくAIが使用されているが、そちらについては「世界はいくつかの悪い出来事が発生することを受け入れるべきだと考えている」とCEOが発言し責任を取る姿勢を全く示さなかったのに対し、こういう時ばかり「今回の脆弱性を発見したのは自… | ↗Not briefed |
arxiv arxiv-ai | 5d | WikiOriginQA: Automated Construction of a QA Dataset with Knowledge Origins for Analyzing Cultural Bias | Existing datasets on cultural knowledge have been typically created by human annotators familiar with each culture. However, the proce… | ↗Not briefed |
arxiv arxiv-ai | 5d | Spectral Smoothing and Spiking Perturbations Reconfigure Capability Profiles and Selective Task Performance in Large Language Models | Singular-value spectra encode important structure in transformer feed-forward projections, yet their causal role in downstream behavio… | ↗Not briefed |
bluesky bluesky-ai | 5d | Funny story: Right now, the LLM end of AI is entirely unprofitable...But the chip makers are making money hand over fist...This dude is doi… | Funny story: Right now, the LLM end of AI is entirely unprofitable...But the chip makers are making money hand over fist...This dude i… | ↗Not briefed |
arxiv arxiv-ai | 5d | RiboBridge: protein-conditioned de novo RNA binder structure design | Designing RNA molecules that bind to target protein structures requires generating both an RNA conformation and its bound complex geom… | ↗Not briefed |
arxiv arxiv-ai | 5d | Fast Iterative Combinatorial Auctions via Deep Reinforcement Learning | Iterative combinatorial auctions are used in spectrum allocation, transportation, and industrial procurement, but rounds are costly. W… | ↗Not briefed |
bluesky bluesky-ai | 5d | A dictionary can, in fact, cost money (see the replies), but also using an LLM to circumvent Finding An Appropriate Resource is very "well … | A dictionary can, in fact, cost money (see the replies), but also using an LLM to circumvent Finding An Appropriate Resource is very "… | ↗Not briefed |
arxiv arxiv-ai | 5d | Compiling Autoregressive LLM Control into Policy-Exact Parallel Inference for Hierarchical RAG | Large language models increasingly act as controllers in structured decision pipelines, yet their decisions are still executed by auto… | ↗Not briefed |
arxiv arxiv-ai | 5d | Group Relative Policy Optimisation Improves Multilingual Speech Recognition in Low-Resource Languages🔧 Whisper | Autoregressive encoder-decoder speech recognition models trained with teacher forcing can suffer from exposure bias, limiting their ab… | ↗Not briefed |
arxiv arxiv-ai | 5d | A Topology-Guided GeoAI Agentic Framework for Urban Flood Screening | Open-world geographic artificial intelligence (GeoAI) agents can transform public geospatial data into accessible decision support for… | ↗Not briefed |
arxiv arxiv-ai | 5d | Anticipating Social Consequences with Distributional World Models | Social world models predict how interactions develop after an action, helping agents negotiate, cooperate and mediate. In these situat… | ↗Not briefed |
arxiv arxiv-ai | 5d | One-shot federated unsupervised domain adaptation with Smoothed Knowledge Distillation and teacher refinement | Federated unsupervised domain adaptation (FUDA) addresses the challenge of adapting models to an unlabeled target domain using decentr… | ↗Not briefed |
arxiv arxiv-ai | 5d | Speculative Acceptance Does Not Decay with Position: A Full-Length Diagnostic on Long Chain-of-Thought with Independent Drafts | The claim that speculative-decoding acceptance rate decays as a chain-of-thought grows longer has become a de facto industry narrative… | ↗Not briefed |
arxiv arxiv-ai | 5d | Asking Less, Adapting Better: Feedback-Efficient Test-Time Adaptation for Vision-and-Language Navigation | Vision-and-Language Navigation (VLN) requires agents to follow instructions and navigate to target locations. However, current VLN age… | ↗Not briefed |
hn hn-ai | 5d | Show HN: Agent Middleware – permits and signed receipts for agent tool calls | — | ↗Not briefed |
hn hn-ai | 5d | I have an issue with what agets affirm, do you have too? | — | ↗Not briefed |
bluesky bluesky-ai | 5d | あれ？１０万社を超えてるのにOpenAIは大赤字なの？？？ なんでかな～！ | あれ？１０万社を超えてるのにOpenAIは大赤字なの？？？ なんでかな～！ | ↗Not briefed |
hn hn-ai | 5d | Shame HN: Distracted Driving on Tesla FSD | I've had Tesla FSD come up in three independent conversations in the last week, as in "whoa isn't it crazy how all thes… | ↗Not briefed |
bluesky bluesky-ai | 5d | literally just load up a deepseek API account with $5 of credits and let it run, it will be enough for the whole thing | literally just load up a deepseek API account with $5 of credits and let it run, it will be enough for the whole thing | ↗Not briefed |
bluesky bluesky-ai | 5d | Sources: DeepSeek raised ~$15B in its latest fundraising round, one of China's largest ever, with CATL and Tencent among the biggest contri… | Sources: DeepSeek raised ~$15B in its latest fundraising round, one of China's largest ever, with CATL and Tencent among the biggest c… | ↗Not briefed |
bluesky bluesky-ai | 5d | カフェでゆっくり過ごすの好き🫶 午後もがんばりましょうね🎶 【 参加者募集中 】 ブログ・日記・イラスト・写真・小説をBlueskyアカウントで投稿・販売できる複合型コンテンツサービス ➡️ https://atelier.fan #ブログ #日記 #イラスト #写真… | カフェでゆっくり過ごすの好き🫶 午後もがんばりましょうね🎶 【 参加者募集中 】 ブログ・日記・イラスト・写真・小説をBlueskyアカウントで投稿・販売できる複合型コンテンツサービス ➡️ https://atelier.fan #ブログ #日記 #イラス… | ↗Not briefed |