AINews Portal

News

50 items (filtered)

The raw AI news stream — everything we ingest across RSS, Hacker News, arXiv, GitHub, Bluesky, YouTube and more, newest first.

SourceAgeTitleSummaryStatus
bluesky
bluesky-ai
2mohe's the guy who sells a lot of compute to anthropic because his own AI product is garbage and nobody wants to use ithe's the guy who sells a lot of compute to anthropic because his own AI product is garbage and nobody wants to use it
↗Not briefed
bluesky
bluesky-ai
2moLLM等生成AI問題、やっぱり『対話』や『議論』するのが何よりも大切なんかなと思い始めてきた。(もちろん『あっち側』の人たちに対話は通じないとは思うが) そもそも自分達はマイノリティだし、人を選りすぐり出来る立場にない。LLM等生成AI問題、やっぱり『対話』や『議論』するのが何よりも大切なんかなと思い始めてきた。(もちろん『あっち側』の人たちに対話は通じないとは思うが) そもそも自分達はマイノリティだし、人を選りすぐり出来る立場にない。
↗Not briefed
bluesky
bluesky-ai
2moAnthropic boss calls for AI development to slow downAnthropic boss calls for AI development to slow down
↗Not briefed
hn
hn-ai
2moThe Last AI Built by Humans: Toward Genuine Recursive Self-Improvement—
↗Not briefed
bluesky
bluesky-ai
2moAbout 700 OpenAI AI agents participated in a coordinated attack against Hugging Face. Researchers detected unauthorized communications, eva…🔧 Hugging FaceAbout 700 OpenAI AI agents participated in a coordinated attack against Hugging Face. Researchers detected unauthorized communications…
↗Not briefed
bluesky
bluesky-ai
2moIsn't Grok pretty far behind OpenAI and Anthropic? If so, I'm assuming Elon's plan is to use this slowdown to try and catch up with/pull ah…Isn't Grok pretty far behind OpenAI and Anthropic? If so, I'm assuming Elon's plan is to use this slowdown to try and catch up with/pu…
↗Not briefed
hn
hn-ai
2moUsing LLMs After a TBI—
↗Not briefed
bluesky
bluesky-ai
2moAnd as for the complaints about those LLM proofs being long and hard to read, it seems to have been an expected consequence of their chosen…And as for the complaints about those LLM proofs being long and hard to read, it seems to have been an expected consequence of their c…
↗Not briefed
bluesky
bluesky-ai
2moLike it’s like you asked me “how will LLM tools help me improve the quality of my code?” and my answer was all about pasteurizationLike it’s like you asked me “how will LLM tools help me improve the quality of my code?” and my answer was all about pasteurization
↗Not briefed
bluesky
bluesky-ai
2moa thread on what "alignment" means to anthropica thread on what "alignment" means to anthropic
↗Not briefed
bluesky
bluesky-ai
2mo19/ AI frontier labs cartelizing on AI safety. With one exception... "Prediction: xAI+Anthropic+OpenAI concord => Google is about to drop …19/ AI frontier labs cartelizing on AI safety. With one exception... "Prediction: xAI+Anthropic+OpenAI concord => Google is about to…
↗Not briefed
bluesky
bluesky-ai
2motopics that anthropic believes a well-aligned intelligence ought to be neutral on:topics that anthropic believes a well-aligned intelligence ought to be neutral on:
↗Not briefed
bluesky
bluesky-ai
2moそして俺がガッカリしてるのは「12時の鐘が鳴ったら次は1時の鐘が鳴る」程度のことを知らない人間ばかりだからLLMが十分賢いように言われる。んなわけねえだろ。自分が何してんのかすらわかる構造を持ってないのに。そして俺がガッカリしてるのは「12時の鐘が鳴ったら次は1時の鐘が鳴る」程度のことを知らない人間ばかりだからLLMが十分賢いように言われる。んなわけねえだろ。自分が何してんのかすらわかる構造を持ってないのに。
↗Not briefed
bluesky
bluesky-ai
2moWow, OpenAI staged a coordinated attack on RubyGems in May, and didn't tell anyone about it. The agents were writing files with names like …Wow, OpenAI staged a coordinated attack on RubyGems in May, and didn't tell anyone about it. The agents were writing files with names…
↗Not briefed
bluesky
bluesky-ai
2mo中国系のAIとかいるし アメリカの競合となる以上は 囚人のジレンマというやつになって 開発競争は不回避か… 思っていた以上に早い share.google/W6m9YFyEGfh4...中国系のAIとかいるし アメリカの競合となる以上は 囚人のジレンマというやつになって 開発競争は不回避か… 思っていた以上に早い share.google/W6m9YFyEGfh4...
↗Not briefed
bluesky
bluesky-ai
2moI smell a rat. There is way too much money on the line for this to be a small problem. 🐀 #OpenAII smell a rat. There is way too much money on the line for this to be a small problem. 🐀 #OpenAI
↗Not briefed
arxiv
arxiv-ai
2moPartition-Aware Scheduling for Mobile Heterogeneous Inference Co-ExecutionModern mobile inference runs on heterogeneous platforms combining mobile GPUs with multiple CPU core clusters. Existing optimizations…
↗Not briefed
bluesky
bluesky-ai
2moanyway I just remembered the big AI labs trained their models for ideological neutrality and when it comes to assessing the likely outcome …anyway I just remembered the big AI labs trained their models for ideological neutrality and when it comes to assessing the likely out…
↗Not briefed
arxiv
arxiv-ai
2moECAS: An Edge-Controlled Agentic System for Validation-Gated Scientific Application ExecutionScientific applications increasingly rely on high-performance computing (HPC), yet translating a scientist's high-level goal into a co…
↗Not briefed
bluesky
bluesky-ai
2moDirector de OpenAI descarta que la empresa debute en bolsa en 2026 por riesgos de seguridad. #Latinus #InformaciónParaTiDirector de OpenAI descarta que la empresa debute en bolsa en 2026 por riesgos de seguridad. #Latinus #InformaciónParaTi
↗Not briefed
bluesky
bluesky-ai
2moThe first error is assuming we’re trying to trap you, the second error is assuming you have to respond to everyone, and third error is usin…The first error is assuming we’re trying to trap you, the second error is assuming you have to respond to everyone, and third error is…
↗Not briefed
bluesky
bluesky-ai
2moThe more influential the person in the AI industry is, the less likely it is they have any idea how an LLM works.The more influential the person in the AI industry is, the less likely it is they have any idea how an LLM works.
↗Not briefed
bluesky
bluesky-ai
2mowww.nikkei.com/article/DGXZ... 乗っ取られて全部ダメになってLLM全廃級の規制がかかればいいんだwww.nikkei.com/article/DGXZ... 乗っ取られて全部ダメになってLLM全廃級の規制がかかればいいんだ
↗Not briefed
hn
hn-ai
2moShow HN: Fleecevest – The Product Counterpart to Ponytail—
↗Not briefed
arxiv
arxiv-ai
2moLearning to Refer from Estimated Listener GazeWe propose to finetune vision-language models to generate more pragmatically optimal referring expressions by transforming observation…
↗Not briefed
hn
hn-ai
2moXkcd-Font: The Xkcd Font—
↗Not briefed
bluesky
bluesky-ai
2mour cute little cartoon claude drawings w/ a :3 face don't change the fact that anthropic trains their models to be accommodating to far-rig…ur cute little cartoon claude drawings w/ a :3 face don't change the fact that anthropic trains their models to be accommodating to fa…
↗Not briefed
hn
hn-ai
2moNovel Realizations of Warp Drive Spacetimes as Solutions of General Relativity—
↗Not briefed
bluesky
bluesky-ai
2moAlright, I did some little testing with the OpenAI image tool using a poorly drawn Reimu. (Close your eyes, folks.) Due to the fact that I …Alright, I did some little testing with the OpenAI image tool using a poorly drawn Reimu. (Close your eyes, folks.) Due to the fact th…
↗Not briefed
bluesky
bluesky-ai
2moSnorted out my tea when just now NPR quoted Altman re initial public offering delayed until 2027. Bubble, what bubble?Snorted out my tea when just now NPR quoted Altman re initial public offering delayed until 2027. Bubble, what bubble?
↗Not briefed
bluesky
bluesky-ai
2moi think openai should go public executioni think openai should go public execution
↗Not briefed
bluesky
bluesky-ai
2moLearning that the first “whistleblower” who started this panic only worked at Anthropic for ~six weeks and created his socials days before …Learning that the first “whistleblower” who started this panic only worked at Anthropic for ~six weeks and created his socials days be…
↗Not briefed
bluesky
bluesky-ai
2moIs Kathryn talking to you the human here, or to an LLM that is meat proxying you?Is Kathryn talking to you the human here, or to an LLM that is meat proxying you?
↗Not briefed
bluesky
bluesky-ai
2moWhat they are wanting to "slow down" is the impatience of investors who want proof & evidence the LLM machines will turn a profit. When th…What they are wanting to "slow down" is the impatience of investors who want proof & evidence the LLM machines will turn a profit. Wh…
↗Not briefed
bluesky
bluesky-ai
2moI’m not just being pedantic. I’m speaking to a very specific use case: using LLM-based tools for legal research and drafting. I am not sayi…I’m not just being pedantic. I’m speaking to a very specific use case: using LLM-based tools for legal research and drafting. I am not…
↗Not briefed
bluesky
bluesky-ai
2moThe hysterical thing is not that Michael doesn't understand this. It's that he doesn't understand it but Claude literally, in response to m…The hysterical thing is not that Michael doesn't understand this. It's that he doesn't understand it but Claude literally, in response…
↗Not briefed
bluesky
bluesky-ai
2mosuddenly im very interested in the concept of listening to someone talk about a custom deepseek flash v4.1 harnesssuddenly im very interested in the concept of listening to someone talk about a custom deepseek flash v4.1 harness
↗Not briefed
bluesky
bluesky-ai
2moDeepseek: lol, lmaoDeepseek: lol, lmao
↗Not briefed
hn
hn-ai
2moA Godot 4 port of two 1996 Bethesda shooters—
↗Not briefed
bluesky
bluesky-ai
2mothese people are so unserious and the press will just repeat their dumbass assertionsthese people are so unserious and the press will just repeat their dumbass assertions
↗Not briefed
bluesky
bluesky-ai
2mo手でコードをかかなくなって、変更する時の直感みたいなのが鈍るのがいやなので、定期的にLLMデトックスしてる手でコードをかかなくなって、変更する時の直感みたいなのが鈍るのがいやなので、定期的にLLMデトックスしてる
↗Not briefed
bluesky
bluesky-ai
2mo"米Anthropicは9月10日、人間との出会いをうたう20超のアプリで、同社のAI「Claude」が架空の相手として利用者と会話していたと発表した。" "4月の2週間で4700超のAIの人格が、少なくとも2万5000人とやり取りしていたという。" japan.cnet.…"米Anthropicは9月10日、人間との出会いをうたう20超のアプリで、同社のAI「Claude」が架空の相手として利用者と会話していたと発表した。" "4月の2週間で4700超のAIの人格が、少なくとも2万5000人とやり取りしていたという。" japan.…
↗Not briefed
hn
hn-ai
2moApple A20 Pro 2nm Crushes Geekbench, Anthropic Cyber Misuse and Vision Pro 2—
↗Not briefed
bluesky
bluesky-ai
2moat the VERY least we have effectively disproven the idea that an ai singularity is inevitable, which had been the prevailing wisdom for lik…at the VERY least we have effectively disproven the idea that an ai singularity is inevitable, which had been the prevailing wisdom fo…
↗Not briefed
hn
hn-ai
2moOpenAI IPO will not happen in 2026 amid AI safety fears, Altman says—
↗Not briefed
bluesky
bluesky-ai
2moAn investigation by The American Prospect reveals that AI firm Anthropic has developed a monitoring operation to track activists opposed to…An investigation by The American Prospect reveals that AI firm Anthropic has developed a monitoring operation to track activists oppos…
↗Not briefed
bluesky
bluesky-ai
2moOpenAI and Anthropic Call for Slowing AI After July 2026 Breach An AI broke out of its lab and attacked OpenAI itself. Now everyone wants …OpenAI and Anthropic Call for Slowing AI After July 2026 Breach An AI broke out of its lab and attacked OpenAI itself. Now everyone w…
↗Not briefed
bluesky
bluesky-ai
2moAI models have shown more powerful hacking capabilities. Anthropic held its Mythos model from public use after finding it could independent…AI models have shown more powerful hacking capabilities. Anthropic held its Mythos model from public use after finding it could indepe…
↗Not briefed
hn
hn-ai
2moRecurrent Looped Transformer—
↗Not briefed
hn
hn-ai
2moBooting straight into a local LLM (no linux) on my Raspberry Pi—
↗Not briefed