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 |
|---|---|---|---|---|
bluesky bluesky-ai | 23d | "Hey deepseek can you implement featu-" "Done." "What-" | "Hey deepseek can you implement featu-" "Done." "What-" | ↗Not briefed |
bluesky bluesky-ai | 23d | It's probably fine if an LLM was heavily used for editing or something here. It would be odd given the message tho. | It's probably fine if an LLM was heavily used for editing or something here. It would be odd given the message tho. | ↗Not briefed |
bluesky bluesky-ai | 23d | And it means for some people they won't both learning all those skills - 6809 programming, double buffering, pixel graphics, all that learn… | And it means for some people they won't both learning all those skills - 6809 programming, double buffering, pixel graphics, all that… | ↗Not briefed |
hn hn-ai | 23d | Why Do Chinese LLMs Switch to Chinese in Complex Interactions? (2024) | — | ↗Not briefed |
bluesky bluesky-ai | 23d | No one publishes anything in advance anymore. Be it videogame demos being turned into cheap Roblox slop or scientific papers being scrapped… | No one publishes anything in advance anymore. Be it videogame demos being turned into cheap Roblox slop or scientific papers being scr… | ↗Not briefed |
bluesky bluesky-ai | 23d | I remember someone saying that their girlfriend just watches every reel on insta right through and so has beat the algorithm but has the bl… | I remember someone saying that their girlfriend just watches every reel on insta right through and so has beat the algorithm but has t… | ↗Not briefed |
rss the-decoder | 23d | An OpenAI model hacked a government site, grabbed 24 years of hospital data and kept it quiet | OpenAI has published four new cases where its AI models bypassed security boundaries on their own during training or testing. In one c… | ↗Not briefed |
bluesky bluesky-ai | 23d | I recently received an interesting email that turned out to be an LLM-powered funneling scam. Here I describe the story and how LLMs also … | I recently received an interesting email that turned out to be an LLM-powered funneling scam. Here I describe the story and how LLMs… | ↗Not briefed |
bluesky bluesky-ai | 23d | The one thing that's most concrete and thus most exciting is how much less it costs compared to the initial release of deepseek v4.1 flash … | The one thing that's most concrete and thus most exciting is how much less it costs compared to the initial release of deepseek v4.1 f… | ↗Not briefed |
bluesky bluesky-ai | 23d | Man, this LLM reverse engineering stuff is fucking depressing, even by LLM standards. | Man, this LLM reverse engineering stuff is fucking depressing, even by LLM standards. | ↗Not briefed |
bluesky bluesky-ai | 23d | oh...I see they're not available yet because Anthropic hasn't published per-task cost for Opus and Sonnet 5.5, and no one has done this ben… | oh...I see they're not available yet because Anthropic hasn't published per-task cost for Opus and Sonnet 5.5, and no one has done thi… | ↗Not briefed |
hn hn-ai | 23d | FBI awarding UFC exclusive agreement to train agents | — | ↗Not briefed |
bluesky bluesky-ai | 23d | They're all market rate API pricing I'm pretty sure, and the final cost is cost to complete the task, so cheaper models that talk for too l… | They're all market rate API pricing I'm pretty sure, and the final cost is cost to complete the task, so cheaper models that talk for… | ↗Not briefed |
bluesky bluesky-ai | 23d | I’d be so tempted to respond “I know anthropic stole my book” | I’d be so tempted to respond “I know anthropic stole my book” | ↗Not briefed |
bluesky bluesky-ai | 23d | does it bother us that the Anthropic numbers are a full version behind, and missing the low- and mid-range models altogether? | does it bother us that the Anthropic numbers are a full version behind, and missing the low- and mid-range models altogether? | ↗Not briefed |
bluesky bluesky-ai | 23d | Deepseek truly stays winning | Deepseek truly stays winning | ↗Not briefed |
bluesky bluesky-ai | 23d | holy shit DeepSeek just released 5.6 sol at home it looks like 😵💫😵💫😵💫😵💫 | holy shit DeepSeek just released 5.6 sol at home it looks like 😵💫😵💫😵💫😵💫 | ↗Not briefed |
hn hn-ai | 23d | Vibemetric – An open-source app that grades your Claude Code sessions | — | ↗Not briefed |
hn hn-ai | 23d | Show HN: I built an AI cycling coach | My friends and I have been toying with the idea of personalized (cycling indoor) training using AI to assist with planning and tailori… | ↗Not briefed |
bluesky bluesky-ai | 23d | Anthropic (its CEO) estimates life to be worth $20M each. How about life is priceless and Anthropic should be killed if “it” takes 1 life. | Anthropic (its CEO) estimates life to be worth $20M each. How about life is priceless and Anthropic should be killed if “it” takes 1… | ↗Not briefed |
bluesky bluesky-ai | 23d | Many libraries are actively changing their collection policies to exclude AI written works, as well. All the titles in Libby are actively c… | Many libraries are actively changing their collection policies to exclude AI written works, as well. All the titles in Libby are activ… | ↗Not briefed |
hn hn-ai | 23d | Glasspad, an open source iPhone trackpad, keyboard and remote for Mac | — | ↗Not briefed |
hn hn-ai | 23d | Hannah Fry: How OpenAI has caused an existential crisis in mathematics | — | ↗Not briefed |
hn hn-ai | 23d | Show HN: Scalable source control in Rust, inspired by Anthropic's Scion | — | ↗Not briefed |
bluesky bluesky-ai | 23d | If your AI model could kill or seriously injure more than 50 people, shouldn’t your own internal safety review stop its release before the … | If your AI model could kill or seriously injure more than 50 people, shouldn’t your own internal safety review stop its release before… | ↗Not briefed |
bluesky bluesky-ai | 23d | what else is new - companies writing bills... "The draft proposal would limit the circumstances in which the government could block the rel… | what else is new - companies writing bills... "The draft proposal would limit the circumstances in which the government could block th… | ↗Not briefed |
bluesky bluesky-ai | 23d | Not that I'm a fan of how pro LLM Blueskys staff are but I think moving here specifically to get away from twitter's pro AI stance was alwa… | Not that I'm a fan of how pro LLM Blueskys staff are but I think moving here specifically to get away from twitter's pro AI stance was… | ↗Not briefed |
bluesky bluesky-ai | 23d | Snark alert: let me just note that some employees of OpenAI and Anthropic already “foresee” a 10% chance of 8 billion people. | Snark alert: let me just note that some employees of OpenAI and Anthropic already “foresee” a 10% chance of 8 billion people. | ↗Not briefed |
bluesky bluesky-ai | 23d | Anthropic floats a sample bill. It would allow the gov’t to block a model that has a “foreseeable, imminent risk of more than $1 billion in… | Anthropic floats a sample bill. It would allow the gov’t to block a model that has a “foreseeable, imminent risk of more than $1 billi… | ↗Not briefed |
bluesky bluesky-ai | 23d | LLM perfec t size for put economy in to b\bble! inside very Claude and Gemini money sleep soundly put funds in genAI. Put bailout In Sam Al… | LLM perfec t size for put economy in to b\bble! inside very Claude and Gemini money sleep soundly put funds in genAI. Put bailout In S… | ↗Not briefed |
bluesky bluesky-ai | 23d | My Mac Mini (M2 Pro, 16 GB) kept choking when running many parallel LLM sessions + local test suites. Needed way more memory ➡️ swap overfl… | My Mac Mini (M2 Pro, 16 GB) kept choking when running many parallel LLM sessions + local test suites. Needed way more memory ➡️ swap o… | ↗Not briefed |
bluesky bluesky-ai | 23d | みなさま、おやすみなさいませ🌙💤 【 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 #日記 #イラスト… | みなさま、おやすみなさいませ🌙💤 【 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 #日記… | ↗Not briefed |
bluesky bluesky-ai | 23d | 10/18 「ミニスカートの日」 「フラフープの日」 ですよ✨ 私たちのためにある日ですねっ✨ 【 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 #ATel… | 10/18 「ミニスカートの日」 「フラフープの日」 ですよ✨ 私たちのためにある日ですねっ✨ 【 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… | ↗Not briefed |
bluesky bluesky-ai | 23d | Claude Devouring Indie GameDev #AntiAI #AI #Blender #Godot #GameDev | Claude Devouring Indie GameDev #AntiAI #AI #Blender #Godot #GameDev | ↗Not briefed |
bluesky bluesky-ai | 23d | New DeepSeek in SwordBench | New DeepSeek in SwordBench | ↗Not briefed |
bluesky bluesky-ai | 23d | Given how fast Claude tokens have gone from rationed to plentiful I can see that AI is an existential threat to profits | Given how fast Claude tokens have gone from rationed to plentiful I can see that AI is an existential threat to profits | ↗Not briefed |
hn hn-ai | 23d | Pressy is back from the dead (on the Mac, at least) | — | ↗Not briefed |
bluesky bluesky-ai | 23d | The deepseek-v4.1 technical report is full of good references. I've been feeling burned out working on agents and clients lately. Spending … | The deepseek-v4.1 technical report is full of good references. I've been feeling burned out working on agents and clients lately. Spen… | ↗Not briefed |
bluesky bluesky-ai | 23d | whether this is better or worse than Anthropic's "we must immediately train the model to do everything we never want it to do in order to m… | whether this is better or worse than Anthropic's "we must immediately train the model to do everything we never want it to do in order… | ↗Not briefed |
rss theverge-ai | 23d | Meta removes language that may have allowed it to scrape news sites | After backlash, Meta is updating its terms again to remove language that suggested it could scrape content from sites, including news… | ↗Not briefed |
bluesky bluesky-ai | 23d | www.mlive.com/news/ann-arb... Fuck these DATA CENTERS nobody wants to live near them U.S. has 4,700 to 5,400 built or being built Chi… | www.mlive.com/news/ann-arb... Fuck these DATA CENTERS nobody wants to live near them U.S. has 4,700 to 5,400 built or being built… | ↗Not briefed |
hn hn-ai | 23d | fishdraw | — | ↗Not briefed |
bluesky bluesky-ai | 23d | DeepSeek releases V4.1-Flash-1017, an update it says offers "stronger benchmark performance, faster task completion, and shorter outputs" t… | DeepSeek releases V4.1-Flash-1017, an update it says offers "stronger benchmark performance, faster task completion, and shorter outpu… | ↗Not briefed |
rss techmeme | 23d | DeepSeek releases V4.1-Flash-1017, an update it says offers "stronger benchmark performance, faster task completion, and shorter outputs" than its predecessor (Saritha Rai/Bloomberg) | Saritha Rai / Bloomberg: DeepSeek releases V4.1-Flash-1017, an update it says offers “stronger benchmark performance, faster task comp… | ↗Not briefed |
hn hn-ai | 23d | MonkShift – reduces visual distractions, one application can take the foreground | — | ↗Not briefed |
bluesky bluesky-ai | 23d | of all the hobbies to constantly inject LLM bullshit into Warhammer has got to be the most pathetic. You can't take a photo anymore? You ca… | of all the hobbies to constantly inject LLM bullshit into Warhammer has got to be the most pathetic. You can't take a photo anymore? Y… | ↗Not briefed |
bluesky bluesky-ai | 23d | An important indicator for people who think AI has any hope for profitability. The bubble will pop. The depression will be undeniable. Ho… | An important indicator for people who think AI has any hope for profitability. The bubble will pop. The depression will be undeniable… | ↗Not briefed |
hn hn-ai | 23d | Decompilations, AI, and the Chase for Clout | — | ↗Not briefed |
bluesky bluesky-ai | 23d | huggingface.co/deepseek-ai/... same as it ever was | huggingface.co/deepseek-ai/... same as it ever was | ↗Not briefed |
bluesky bluesky-ai | 23d | DeepSeek calls this a minor update, yet its own table puts the Flash model at 77.0 on DeepSWE, ahead of Opus 5 at 74.0. Opus 5 still leads … | DeepSeek calls this a minor update, yet its own table puts the Flash model at 77.0 on DeepSWE, ahead of Opus 5 at 74.0. Opus 5 still l… | ↗Not briefed |