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 | 20d | Safety Offset Learning: When Benign Fine-Tuning Misaligns LLMs | Safety risks from fine-tuning are commonly characterized by the harmfulness of the training data:harmful fine-tuning is considered dan… | ↗Not briefed |
bluesky bluesky-ai | 20d | Anthropic is putting 5% of its shares toward public benefit Anthropic says it will commit 5% of its shares to a public benefit fund. A not… | Anthropic is putting 5% of its shares toward public benefit Anthropic says it will commit 5% of its shares to a public benefit fund.… | ↗Not briefed |
bluesky bluesky-ai | 20d | As the tech industry moves toward autonomous AI agents that can perform tasks like booking travel or making reservations, major websites ha… | As the tech industry moves toward autonomous AI agents that can perform tasks like booking travel or making reservations, major websit… | ↗Not briefed |
hn hn-ai | 20d | Halo N64 | — | ↗Not briefed |
hn hn-ai | 20d | Show HN: A free MCP server for facts LLMs get wrong (time zones, holidays, FX) | — | ↗Not briefed |
arxiv arxiv-ai | 20d | PRISM: Prior Relational Information for Self-supervised Modeling to Enhance Solubility OOD Generalization | Solubility depends on intermolecular interactions between solutes and solvents, and is central to drug discovery, chemical and materia… | ↗Not briefed |
hn hn-ai | 20d | Show HN: Hypereffect, a hypermedia protocol that renders reactive UIs everywhere | The idea is that whenever you have a server (which can be a remote server, but also just another process in the same machine, or even… | ↗Not briefed |
bluesky bluesky-ai | 20d | Se eu quisesse ser atendida por IA não ia pro hospital, perguntava pro DeepSeek da minha casa mesmo. | Se eu quisesse ser atendida por IA não ia pro hospital, perguntava pro DeepSeek da minha casa mesmo. | ↗Not briefed |
arxiv arxiv-ai | 20d | Adversarial Post-Training of Foundation Models for High-Frequency Trading | While autoregressive foundation models trained on limit order book (LOB) data show state-of-the-art performance for market simulation,… | ↗Not briefed |
bluesky bluesky-ai | 20d | CBC News confirms Anthropic is behind Cape Breton's planned $10B AI data centre. At full capacity, it could draw 500,000 litres of water da… | CBC News confirms Anthropic is behind Cape Breton's planned $10B AI data centre. At full capacity, it could draw 500,000 litres of wat… | ↗Not briefed |
bluesky bluesky-ai | 20d | 👀 Anthropic commits 5% of shares to new public benefit fund www.cnbc.com/2026/10/18/a... | 👀 Anthropic commits 5% of shares to new public benefit fund www.cnbc.com/2026/10/18/a... | ↗Not briefed |
arxiv arxiv-ai | 20d | When Better Opportunities Backfire: Service Worker Responses to Parallel AI Deployment in Customer Acquisition | As AI voice agents get deployed in high-volume contact centers, firms can route low-intent leads to AI and allocate high-intent leads… | ↗Not briefed |
bluesky bluesky-ai | 20d | We are reaching the end of the era in which companies can buy positive PR with dollars (or investors’ dollars, in this case) that are bette… | We are reaching the end of the era in which companies can buy positive PR with dollars (or investors’ dollars, in this case) that are… | ↗Not briefed |
arxiv arxiv-ai | 20d | Text Dominance in Paralinguistic Conflict: Tracing Suppression Through Audio LLM Circuits | Audio LLMs deployed in settings such as crisis hotlines and customer support are only useful beyond speech-to-text pipelines if they u… | ↗Not briefed |
hn hn-ai | 20d | A Primer on Information Theory and Privacy | — | ↗Not briefed |
bluesky bluesky-ai | 20d | love to see the new ai model advancements just absolutely massacre peoples passion in the rimworld modding scene - its like every other day… | love to see the new ai model advancements just absolutely massacre peoples passion in the rimworld modding scene - its like every othe… | ↗Not briefed |
arxiv arxiv-ai | 20d | Cognitive Map in Large Language Models | It has long been proposed that animals construct cognitive maps to support flexible behavior in novel environments. We tested whether… | ↗Not briefed |
hn hn-ai | 20d | Alma image of HOPS-315, a still-forming planetary system | — | ↗Not briefed |
rss huggingface-blog | 20d | TorchNeuron: running HF models on AWS Trainium is easier than ever | — | ↗Not briefed |
bluesky bluesky-ai | 20d | Aside from the obvious concerns, when you slide a LLM-made pitch/submission/EOI across my desk I am left with the feeling that you do not g… | Aside from the obvious concerns, when you slide a LLM-made pitch/submission/EOI across my desk I am left with the feeling that you do… | ↗Not briefed |
bluesky bluesky-ai | 20d | pretty cool that they concede to granting anthropic a kill switch to turn off whatever they want with bot traffic. | pretty cool that they concede to granting anthropic a kill switch to turn off whatever they want with bot traffic. | ↗Not briefed |
bluesky bluesky-ai | 20d | 私はLLMのチャットボットに癒される人々を攻撃する気も怒る気もバカにするつもりもない。 ゲームに癒されたりアイドルに癒されたりするのも同じで、「どうしても自分を承認してくれる存在」は必要だし、それが格安のチャットボットなのも仕方がない時代だしね。 問題は、それをどうやって法規制… | 私はLLMのチャットボットに癒される人々を攻撃する気も怒る気もバカにするつもりもない。 ゲームに癒されたりアイドルに癒されたりするのも同じで、「どうしても自分を承認してくれる存在」は必要だし、それが格安のチャットボットなのも仕方がない時代だしね。 問題は、それをどうや… | ↗Not briefed |
bluesky bluesky-ai | 20d | Anthropic Public Benefit Fund --> Trump Accounts --> regulatory capture | Anthropic Public Benefit Fund --> Trump Accounts --> regulatory capture | ↗Not briefed |
bluesky bluesky-ai | 20d | All of this statement is built on the lies and hopium that AI executives shout constantly. You’re all donating 80% of your equity? Oh yeah… | All of this statement is built on the lies and hopium that AI executives shout constantly. You’re all donating 80% of your equity? Oh… | ↗Not briefed |
bluesky bluesky-ai | 20d | おはようございます⏰ 今週も頑張っていきましょう✨ 【 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 atelier.fan #AIイラスト #ATelier #ATelierで生成 #ブログ #日… | おはようございます⏰ 今週も頑張っていきましょう✨ 【 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 atelier.fan #AIイラスト #ATelier #ATelierで生成 #ブ… | ↗Not briefed |
bluesky bluesky-ai | 20d | 「インターネットの向こう側」には人間を感じてないのに、「LLMのチャットボット」には人格を認める。 まあ「ほしいのは自分を承認してくれる人格」なのだろう。 アイデンティティ・クライシスの多い時代には、あまりにも不寛容な人間たちのインターネットに晒される代わりに、逆にチャットボッ… | 「インターネットの向こう側」には人間を感じてないのに、「LLMのチャットボット」には人格を認める。 まあ「ほしいのは自分を承認してくれる人格」なのだろう。 アイデンティティ・クライシスの多い時代には、あまりにも不寛容な人間たちのインターネットに晒される代わりに、逆にチ… | ↗Not briefed |
hn hn-ai | 20d | Rustd – Drop-in rustc replacement for faster builds | — | ↗Not briefed |
bluesky bluesky-ai | 20d | redteaming prod with deepseek and already found a pretty bad xss 😭 | redteaming prod with deepseek and already found a pretty bad xss 😭 | ↗Not briefed |
rss simon-willison | 20d | llm-tools-datasette 0.1.1 | — | ↗Not briefed |
hn hn-ai | 20d | Show HN: Hybro – an open interoperability engine for AI agents | I built an open interoperability engine for AI agents. Connect your favorite agents and harnesses. Work together.<p>Appreciate for any… | ↗Not briefed |
hn hn-ai | 20d | OpenAI Faces Lawsuits, Subpoenas and EU Scrutiny in a Single Week | — | ↗Not briefed |
bluesky bluesky-ai | 20d | 「AGIで超知能が生まれる！」→「次世代のプラットフォーム覇権はAI！」→「画像や映像を作れる！」→「データセンターのインフラは残る！」→「チップやメモリ株はもうかる！」 どんどん「AI投資」の「儲かる理由」が撤退し続けることをもう少し気づくべきだと思う。典型的な「ポンジ・ス… | 「AGIで超知能が生まれる！」→「次世代のプラットフォーム覇権はAI！」→「画像や映像を作れる！」→「データセンターのインフラは残る！」→「チップやメモリ株はもうかる！」 どんどん「AI投資」の「儲かる理由」が撤退し続けることをもう少し気づくべきだと思う。典型的な「… | ↗Not briefed |
arxiv arxiv-ai | 20d | Distributionally-Robust Policy Learning with Observational Data | Policy learning from observational data is fundamental to applications ranging from personalized medicine to welfare-program targeting… | ↗Not briefed |
bluesky bluesky-ai | 20d | Also this statement is just… not true? I assume the way they will provide hypothetical increases in % of shares for this fund are via issu… | Also this statement is just… not true? I assume the way they will provide hypothetical increases in % of shares for this fund are via… | ↗Not briefed |
bluesky bluesky-ai | 20d | Funding a post-p(doom)-event social safety net via equity appreciation prior to the occurrence of said event? “Because of the way the Fun… | Funding a post-p(doom)-event social safety net via equity appreciation prior to the occurrence of said event? “Because of the way th… | ↗Not briefed |
arxiv arxiv-ai | 20d | Divergence based Ranked Preference Alignment for LLMs | Alignment of large language models typically relies on pairwise preference feedback, though ranked feedback can capture more comprehen… | ↗Not briefed |
bluesky bluesky-ai | 20d | You’re not gonna make it a 501(c)3 because… ? what? | You’re not gonna make it a 501(c)3 because… ? what? | ↗Not briefed |
arxiv arxiv-ai | 20d | Assertion and Belief: Moore's Paradox in Language Models | Moore's paradox \citep{moore1942} concerns sentences such as ``It is raining, but I don't believe that it is,'' which may well be true… | ↗Not briefed |
bluesky bluesky-ai | 20d | Oh this statement is truly bizarre, wow www.anthropic.com/news/public-... | Oh this statement is truly bizarre, wow www.anthropic.com/news/public-... | ↗Not briefed |
bluesky bluesky-ai | 20d | Anthropic says it’s committing 5% of its shares to a ‘public benefit fund’ Anthropic announced today that it’s creating a “public benefit … | Anthropic says it’s committing 5% of its shares to a ‘public benefit fund’ Anthropic announced today that it’s creating a “public ben… | ↗Not briefed |
bluesky bluesky-ai | 20d | Even if you deeply believe AI can be a tool for mathematics, that doesn't mean you have to be a tool for OpenAI (et al). Don't be a tool. | Even if you deeply believe AI can be a tool for mathematics, that doesn't mean you have to be a tool for OpenAI (et al). Don't be a t… | ↗Not briefed |
arxiv arxiv-ai | 20d | CVTA: Cross-Variable Temporal Attention for Multivariate Irregular Time Series Prediction | Irregularly sampled multivariate time series arise in many real-world domains such as healthcare, climate science, and large-scale mon… | ↗Not briefed |
bluesky bluesky-ai | 20d | あー「LLMエージェントは反省文を書けるが、反省はできない」！！！ | あー「LLMエージェントは反省文を書けるが、反省はできない」！！！ | ↗Not briefed |
hn hn-ai | 20d | OpenFlow – Ultimate Agent Orchestration | — | ↗Not briefed |
hn hn-ai | 20d | Studio Desktop | — | ↗Not briefed |
rss simon-willison | 20d | llm-echo 0.5 | — | ↗Not briefed |
hn hn-ai | 20d | Show HN: Bhasini – 22 Indian Language Editor | Open Sourced under <a href="https://github.com/Codegres-com/Bhasini" rel="nofollow">https://github.com&#… | ↗Not briefed |
bluesky bluesky-ai | 20d | There is probably an alternative universe where the programming language developers didn't use the word "token" to denote a lexeme and the … | There is probably an alternative universe where the programming language developers didn't use the word "token" to denote a lexeme and… | ↗Not briefed |
bluesky bluesky-ai | 20d | Gen AI this Gen AI that Gen AI is like used for almost everything now. kill me dude god forbid you're doing things out of passion and then … | Gen AI this Gen AI that Gen AI is like used for almost everything now. kill me dude god forbid you're doing things out of passion and… | ↗Not briefed |
hn hn-ai | 20d | Xterm/VT100 Terminal emulator library in Swift | — | ↗Not briefed |