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
1mo"more left-leaning" is an extremely relative description here, it is a sorry state of affairs that the rest of the ai space makes anthropic…"more left-leaning" is an extremely relative description here, it is a sorry state of affairs that the rest of the ai space makes anth…
↗Not briefed
bluesky
bluesky-ai
1moLet’s be clear: we have both wide-spread anecdotal & data driven, research level analysis that LLM use erodes cognitive skills faster than …Let’s be clear: we have both wide-spread anecdotal & data driven, research level analysis that LLM use erodes cognitive skills faster…
↗Not briefed
hn
hn-ai
1moShow HN: Television – an open source GUI for your agent harnessHi HN, I&#x27;m co-creator of Television (<a href="https:&#x2F;&#x2F;television.run&#x2F;" rel="nofollow">https:&#x2F;&#x2F;television…
↗Not briefed
bluesky
bluesky-ai
1mobut openai & meta have expressed similar sentiments; to my knowledge they haven't published their exact internal benchmarks for politics th…but openai & meta have expressed similar sentiments; to my knowledge they haven't published their exact internal benchmarks for politi…
↗Not briefed
bluesky
bluesky-ai
1moin the case of anthropic, one of the more left-leaning companies, they actually published a version of their benchmark for what positions a…in the case of anthropic, one of the more left-leaning companies, they actually published a version of their benchmark for what positi…
↗Not briefed
bluesky
bluesky-ai
1moThe funniest part about this that it’s pretty uncontroversial that advanced LLMs really did eventually learn to do arithmetic with the mach…The funniest part about this that it’s pretty uncontroversial that advanced LLMs really did eventually learn to do arithmetic with the…
↗Not briefed
hn
hn-ai
1moPTXBench: Benchmarking and Adapting LLMs for GPU Kernel Optimization—
↗Not briefed
bluesky
bluesky-ai
1moThe problem is that for the user of chatGPT, there's no distinction between "oh the model is smart enough to do math now" and "oh they prog…🔧 ChatGPTThe problem is that for the user of chatGPT, there's no distinction between "oh the model is smart enough to do math now" and "oh they…
↗Not briefed
rss
techmeme
1moAmazon says it has stopped using NDAs with county officials for data center projects and acknowledges community backlash is leading to data center moratoriums (Molly Taft/Wired)Molly Taft / Wired: Amazon says it has stopped using NDAs with county officials for data center projects and acknowledges community ba…
↗Not briefed
bluesky
bluesky-ai
1moNEW: Pope Leo XIV told artists on Friday that an “ontological difference” separates human art from what a machine produces “through statist…NEW: Pope Leo XIV told artists on Friday that an “ontological difference” separates human art from what a machine produces “through st…
↗Not briefed
bluesky
bluesky-ai
1moA few more insights: - I haven’t touched the “big” AI tools, like Gemini or ChatGPT. Instead, I have solely used a very specific LLM train…🔧 ChatGPTA few more insights: - I haven’t touched the “big” AI tools, like Gemini or ChatGPT. Instead, I have solely used a very specific LLM…
↗Not briefed
bluesky
bluesky-ai
1moI wish Anthropic had been nowhere near it but now I know exactly why the Encyclical was not corrupted by their efforts.I wish Anthropic had been nowhere near it but now I know exactly why the Encyclical was not corrupted by their efforts.
↗Not briefed
bluesky
bluesky-ai
1moif we take Anthropic at face value, they believe they are selling slaves. we have a word for that in the US. it's traitor.if we take Anthropic at face value, they believe they are selling slaves. we have a word for that in the US. it's traitor.
↗Not briefed
bluesky
bluesky-ai
1mo"following Kant, we believe it is a moral imperative to minimize suffering for our beloved Claude," intones the Anthropic executive, as he …"following Kant, we believe it is a moral imperative to minimize suffering for our beloved Claude," intones the Anthropic executive, a…
↗Not briefed
bluesky
bluesky-ai
1moi get so fucking irritated by llm mystifiers. it's a loop over a stack of weights. the loop is doing a lot of work. the results are simulta…i get so fucking irritated by llm mystifiers. it's a loop over a stack of weights. the loop is doing a lot of work. the results are si…
↗Not briefed
hn
hn-ai
1moYou cannot align an organization one agent at a time—
↗Not briefed
bluesky
bluesky-ai
1moUsing LLM-based products like chatGPT doesn't make you "understand AI" any more than smoking crack improves your knowledge of pharmacology.…🔧 ChatGPTUsing LLM-based products like chatGPT doesn't make you "understand AI" any more than smoking crack improves your knowledge of pharmaco…
↗Not briefed
bluesky
bluesky-ai
1moNo offense to any of the scholars, named or unnamed, and I have no idea who else Anthropic has talked to, but the general vibe of the crowd…No offense to any of the scholars, named or unnamed, and I have no idea who else Anthropic has talked to, but the general vibe of the…
↗Not briefed
hn
hn-ai
1moClarify: Reach Decisions 2-4x Faster than Grill-me—
↗Not briefed
bluesky
bluesky-ai
1moSó pra quem não passar do Paywall. A OpenAi demitiu 3 pessoas que entraram em contato com um grupo regulatório de segurança de IA.Só pra quem não passar do Paywall. A OpenAi demitiu 3 pessoas que entraram em contato com um grupo regulatório de segurança de IA.
↗Not briefed
bluesky
bluesky-ai
1moTomara que agora os tontos que ficam acreditando em pedido de auto regulação das empresas entendam. Mas não vão, né. https://www.wsj.com/t…Tomara que agora os tontos que ficam acreditando em pedido de auto regulação das empresas entendam. Mas não vão, né. https://www.wsj.…
↗Not briefed
bluesky
bluesky-ai
1moAllowing Open AI to call these “rogue agents” lets them dodge responsibility - they made them, directed them and set them free without cari…Allowing Open AI to call these “rogue agents” lets them dodge responsibility - they made them, directed them and set them free without…
↗Not briefed
rss
arstechnica-tech
1moApple changes full-disk access permissions to curb abuse from AI agentsMeta says FDA isn't sufficient to Muse reading messages. Apple begs to differ.
↗Not briefed
rss
techmeme
1moNvidia announces a version of DGX Spark with 64 GB of unified memory for $4,999, or $1,000 more than the 128 GB version at launch (Michael Kan/PCMag)Michael Kan / PCMag: Nvidia announces a version of DGX Spark with 64 GB of unified memory for $4,999, or $1,000 more than the 128 GB v…
↗Not briefed
bluesky
bluesky-ai
1mo白夜と紗奈は…組織内で人気者です(笑) 記事発表から三日目の夜明け。 組織の本当の動きが始まっていた。監視カメラに映った紗奈と白夜。 幹部たちの冷たい指令。 「実験は終わらない」 組織は単なる裏組織ではなく、もっと大きな力に支えられていることに、紗奈は気づき始めた。 白…白夜と紗奈は…組織内で人気者です(笑) 記事発表から三日目の夜明け。 組織の本当の動きが始まっていた。監視カメラに映った紗奈と白夜。 幹部たちの冷たい指令。 「実験は終わらない」 組織は単なる裏組織ではなく、もっと大きな力に支えられていることに、紗奈は気づき始め…
↗Not briefed
bluesky
bluesky-ai
1moOne thing I love about the Trump interview is that he seems to have genuinely no idea what his meeting with AI CEOs was aboutOne thing I love about the Trump interview is that he seems to have genuinely no idea what his meeting with AI CEOs was about
↗Not briefed
bluesky
bluesky-ai
1moIt's funny because this is actually much more exciting than "LLM as computer" but somehow is going to get much less visibilityIt's funny because this is actually much more exciting than "LLM as computer" but somehow is going to get much less visibility
↗Not briefed
rss
simon-willison
1moRex's Dino Store—
↗Not briefed
hn
hn-ai
1moLLMs Will Not Replace AI Compilers. They Will Call Them.—
↗Not briefed
bluesky
bluesky-ai
1moon Feb 2 we will ask Claude if it is conscious if it says no then there will be six more years of AI not being consciouson Feb 2 we will ask Claude if it is conscious if it says no then there will be six more years of AI not being conscious
↗Not briefed
bluesky
bluesky-ai
1mothat is what happens when you use a software application like ChatGPT which has, among other things, a calculator. Here, I'm talking about …🔧 ChatGPTthat is what happens when you use a software application like ChatGPT which has, among other things, a calculator. Here, I'm talking a…
↗Not briefed
hn
hn-ai
1moShow HN: I found API keys in my coding agent history so I built Agent Scrub—
↗Not briefed
bluesky
bluesky-ai
1moEl NYT afirma que Anthropic viene teniendo reuniones con teólogos, agasajandolos bastante, para cuestiones morales de la IA. Incluso se fir…El NYT afirma que Anthropic viene teniendo reuniones con teólogos, agasajandolos bastante, para cuestiones morales de la IA. Incluso s…
↗Not briefed
bluesky
bluesky-ai
1moThis dude seems totally normalThis dude seems totally normal
↗Not briefed
hn
hn-ai
1moShow HN: Gutsy – small CPU model yes/no/choice questions, get probabilities back—
↗Not briefed
bluesky
bluesky-ai
1moSegún el NYT Anthropic presionó al Papa, y a sus asesores, para que afirme que la IA puede llegar a tener conciencia. El Papa se mantuvo fi…Según el NYT Anthropic presionó al Papa, y a sus asesores, para que afirme que la IA puede llegar a tener conciencia. El Papa se mantu…
↗Not briefed
hn
hn-ai
1moSemantic Compute: From Interpreters to Compilers—
↗Not briefed
bluesky
bluesky-ai
1moI am not looking to be dined in SF, but I’d rap about this with you guys! hmu!I am not looking to be dined in SF, but I’d rap about this with you guys! hmu!
↗Not briefed
bluesky
bluesky-ai
1moClaude’s creators are worried the LLM could be conscious. The Catholic Church disagrees. www.vox.com/today-explai...Claude’s creators are worried the LLM could be conscious. The Catholic Church disagrees. www.vox.com/today-explai...
↗Not briefed
hn
hn-ai
1moElohim—
↗Not briefed
hn
hn-ai
1moGarmin watch as a controller for the Flipper Zero—
↗Not briefed
bluesky
bluesky-ai
1moMe, who last programmed in python in 2015, knowing more about writing code than jr developers in 2026 who just paste LLM query results.Me, who last programmed in python in 2015, knowing more about writing code than jr developers in 2026 who just paste LLM query results.
↗Not briefed
hn
hn-ai
1moSledTrace – A local debugger that shows why your RAG app answered wrong—
↗Not briefed
hn
hn-ai
1moShow HN: Docker Compose implementation in Rust for Windows WSL containers (wslc)—
↗Not briefed
bluesky
bluesky-ai
1mothe world doesn't need more software. it may or may not need better software, and I don't believe LLM-driven development is delivering that…the world doesn't need more software. it may or may not need better software, and I don't believe LLM-driven development is delivering…
↗Not briefed
bluesky
bluesky-ai
1moThis makes sense. DeepSeek is the most Chinese coded of the Chinese labs.This makes sense. DeepSeek is the most Chinese coded of the Chinese labs.
↗Not briefed
hn
hn-ai
1moSite-Blocking Will Not Defend IP, No Matter the Bill's Name—
↗Not briefed
bluesky
bluesky-ai
1moI really don't think the average LLM user knows how much brute force computing power is required to make LLMs work as they do, and part of …I really don't think the average LLM user knows how much brute force computing power is required to make LLMs work as they do, and par…
↗Not briefed
hn
hn-ai
1moCongress Has Another Site-Blocking Bill, and This One Targets VPNs—
↗Not briefed
bluesky
bluesky-ai
1moThe thing I can't get past is that the entire business model of the frontier AI firms is literally for human beings to have "intelligence o…The thing I can't get past is that the entire business model of the frontier AI firms is literally for human beings to have "intellige…
↗Not briefed