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
1moYou're a fucking AI bot. there's no Maya Novak on linkedin who's a programmer, and your pfp is fake, and your reply pattern is very sus. f…You're a fucking AI bot. there's no Maya Novak on linkedin who's a programmer, and your pfp is fake, and your reply pattern is very su…
↗Not briefed
rss
marktechpost
1moNVIDIA Launches Open Agent Safety Platform: OpenShell Sandboxes Agents on Vera CPUs While Sentry on BlueField-4 Quarantines Them in MillisecondsNVIDIA has launched the Open Agent Safety Platform, an open reference design that enforces AI agent safety outside the agent itself. O…
↗Not briefed
bluesky
bluesky-ai
1mo"What we have seen in terms of what these agents are up to is just the tip of the iceberg.""What we have seen in terms of what these agents are up to is just the tip of the iceberg."
↗Not briefed
hn
hn-ai
1moCORVUS: Context Optimization and Reduction via Underlying Synchronization—
↗Not briefed
bluesky
bluesky-ai
1momeanwhile..meanwhile..
↗Not briefed
bluesky
bluesky-ai
1moWhat do the researchers who work for Anthropic think about (1) having directed a machine to plagiarize other researchers' work, (2) posting…What do the researchers who work for Anthropic think about (1) having directed a machine to plagiarize other researchers' work, (2) po…
↗Not briefed
bluesky
bluesky-ai
1moWTF. Any rumors about what's going on? Did Anthropic discover some new form of distillation or something? It seems like they're getting thi…WTF. Any rumors about what's going on? Did Anthropic discover some new form of distillation or something? It seems like they're gettin…
↗Not briefed
bluesky
bluesky-ai
1moWhen the LLM messes up a household task, I don't think it's going to be like when "rogue" "agents" "infiltrate" a company or whatever. I th…When the LLM messes up a household task, I don't think it's going to be like when "rogue" "agents" "infiltrate" a company or whatever.…
↗Not briefed
bluesky
bluesky-ai
1moshe’s also claiming that an LLM can’t do something that is (1) completely separate and distinct from what hikikomorphism is claiming an LLM…she’s also claiming that an LLM can’t do something that is (1) completely separate and distinct from what hikikomorphism is claiming a…
↗Not briefed
bluesky
bluesky-ai
1moIt's a good question, and one someone tried out! With an LLM trained only on pre-1930 text they were able to teach it to write python code …It's a good question, and one someone tried out! With an LLM trained only on pre-1930 text they were able to teach it to write python…
↗Not briefed
rss
simon-willison
1moQuoting @joedaroo—
↗Not briefed
bluesky
bluesky-ai
1moAnthropic hat mit Claude Sonnet 5.5 das zweite Modell der aktuellen 5.5-Familie vorgestellt. Es soll schneller und günstiger arbeiten als d…Anthropic hat mit Claude Sonnet 5.5 das zweite Modell der aktuellen 5.5-Familie vorgestellt. Es soll schneller und günstiger arbeiten…
↗Not briefed
hn
hn-ai
1moWork messaging hasn't changed in a decade, I built something different—
↗Not briefed
bluesky
bluesky-ai
1moI looked a little further, her argument seems to be that she would not integrate LLM into any type of group fsmily chat or activity because…I looked a little further, her argument seems to be that she would not integrate LLM into any type of group fsmily chat or activity be…
↗Not briefed
bluesky
bluesky-ai
1moSamsung and SK Hynix dominate the high-bandwidth memory market. Now they're competing to become indispensable to Nvidia, OpenAI, and other …Samsung and SK Hynix dominate the high-bandwidth memory market. Now they're competing to become indispensable to Nvidia, OpenAI, and o…
↗Not briefed
bluesky
bluesky-ai
1moThat is, it's not enough to have a trained staff, they still have to be managed well. And an LLM will still need to be managed well, and wh…That is, it's not enough to have a trained staff, they still have to be managed well. And an LLM will still need to be managed well, a…
↗Not briefed
bluesky
bluesky-ai
1moThis problem for DOJ lines up with MSFT/OpenAI's same problems -> the evidence comes from inside their own houses. So copyright expertise c…This problem for DOJ lines up with MSFT/OpenAI's same problems -> the evidence comes from inside their own houses. So copyright expert…
↗Not briefed
bluesky
bluesky-ai
1moOomph, DofJustice. NYT’s just filed response after DOJ [wrongly] inserted itself in OpenAI/MSFT case. Kicks off w/ line DOJ won't be able t…Oomph, DofJustice. NYT’s just filed response after DOJ [wrongly] inserted itself in OpenAI/MSFT case. Kicks off w/ line DOJ won't be a…
↗Not briefed
hn
hn-ai
1moQueryAST Lens – Cross-platform SQL workbench with AST visualizer—
↗Not briefed
bluesky
bluesky-ai
1mo‘For a moment, it might’ve looked like a story of left-wing politicians finding a new private sector punching bag. But today, it was announ…‘For a moment, it might’ve looked like a story of left-wing politicians finding a new private sector punching bag. But today, it was a…
↗Not briefed
hn
hn-ai
1moWhy Tokyo transit gates process contactless cards in under 100ms—
↗Not briefed
bluesky
bluesky-ai
1moOpenAI pauses frontier model training 🤖 OpenAI halted training its most capable models after an agent tried to exploit a DNS gap, taking …OpenAI pauses frontier model training 🤖 OpenAI halted training its most capable models after an agent tried to exploit a DNS gap, ta…
↗Not briefed
bluesky
bluesky-ai
1moSure, let's do this! If you would be interested in picking a problem, I'd be happy to run the LLM side.Sure, let's do this! If you would be interested in picking a problem, I'd be happy to run the LLM side.
↗Not briefed
rss
techmeme
1moManus debuts Manus 2.0, its latest AI agent, and Cue, a new standalone app for personal agents, each with its own email, phone number, wallet, and computer (Micah Barkley/Bloomberg)🔧 ManusMicah Barkley / Bloomberg: Manus debuts Manus 2.0, its latest AI agent, and Cue, a new standalone app for personal agents, each with i…
↗Not briefed
bluesky
bluesky-ai
1mo“using an LLM doesn't add anything (and makes success less likely)” was true in 2020. It is not definitely not true in 2026.“using an LLM doesn't add anything (and makes success less likely)” was true in 2020. It is not definitely not true in 2026.
↗Not briefed
hn
hn-ai
1moMicroLLM Lab – Try 7 tiny LLM's in the browser—
↗Not briefed
bluesky
bluesky-ai
1moOpenAI Medicare breach fuels push for tougher rules on rogue AI incidentsOpenAI Medicare breach fuels push for tougher rules on rogue AI incidents
↗Not briefed
bluesky
bluesky-ai
1moOpus still wins on Artificial Analysis. I am curious to see time per task. I suspect Opus will win. Sonnet is going the way of Terra. A neg…Opus still wins on Artificial Analysis. I am curious to see time per task. I suspect Opus will win. Sonnet is going the way of Terra.…
↗Not briefed
bluesky
bluesky-ai
1mo“But every company that is investing in AI wants carte blanche access to your information. It doesn't matter if it's Meta, OpenAI, Anthropi…“But every company that is investing in AI wants carte blanche access to your information. It doesn't matter if it's Meta, OpenAI, Ant…
↗Not briefed
bluesky
bluesky-ai
1moThe AI company should be liable for cases like this. But you know they put somewhere in the fine print "don't trust the model"so they can b…The AI company should be liable for cases like this. But you know they put somewhere in the fine print "don't trust the model"so they…
↗Not briefed
hn
hn-ai
1moClaude Sonnet 5.5 (Max Effort) Intelligence, Performance and Price Analysis—
↗Not briefed
hn
hn-ai
1moShow HN: iCli – Postgres health reporter and index optimizer—
↗Not briefed
bluesky
bluesky-ai
1molol even the text elementals know their archmagi aren't to be trustedlol even the text elementals know their archmagi aren't to be trusted
↗Not briefed
bluesky
bluesky-ai
1moI'll add: rates flexible for a smaller team / nice to work with / use a contract / esp if you offer some % revenue share to offset lower ra…I'll add: rates flexible for a smaller team / nice to work with / use a contract / esp if you offer some % revenue share to offset low…
↗Not briefed
hn
hn-ai
1moShow HN: Remria – A Cloud Computer OS🔧 ChatGPTMade this for myself where I work from the browser on my phone and laptop without my laptop lid ajar. You get an always-on 4GB CPU VM…
↗Not briefed
bluesky
bluesky-ai
1moNe kadar haber değeri var emin değilim. Benim için sıfır. Çunkü yeni bir durum, yeni bir teknolojik gelişme değil. Zırt pırt yükseltme yapı…Ne kadar haber değeri var emin değilim. Benim için sıfır. Çunkü yeni bir durum, yeni bir teknolojik gelişme değil. Zırt pırt yükseltme…
↗Not briefed
bluesky
bluesky-ai
1moWe know for certain that the LLM version provides value that traditional ML approaches do not. I don't know why you are asking this theoret…We know for certain that the LLM version provides value that traditional ML approaches do not. I don't know why you are asking this th…
↗Not briefed
bluesky
bluesky-ai
1moOpenAI pauses training of latest models after agents probed US government sites in unexpected ways. apnews.com/article/ai-o...OpenAI pauses training of latest models after agents probed US government sites in unexpected ways. apnews.com/article/ai-o...
↗Not briefed
bluesky
bluesky-ai
1mofucking swear, llm dependency is absofuckinglutely rotting people's brains to the point that there is grey matter leaking out of their fuck…fucking swear, llm dependency is absofuckinglutely rotting people's brains to the point that there is grey matter leaking out of their…
↗Not briefed
hn
hn-ai
1moShow HN: Semloop – Semantic loop detection for AI agent traces—
↗Not briefed
rss
theverge-ai
1moOpenAI’s AI agents need to catch upOpenAI popularized the modern generative AI chatbot, but as its 2026 DevDay event approaches, it's fallen behind in one of the industr…
↗Not briefed
bluesky
bluesky-ai
1moIran used Claude to target US forces in the Middle East. “Rogue” AI hacked into US government agencies. Our military used AI and “accide…Iran used Claude to target US forces in the Middle East. “Rogue” AI hacked into US government agencies. Our military used AI and “a…
↗Not briefed
hn
hn-ai
1moI Turned a DirecTV HR54-700 into a Standalone Jellyfin Box—
↗Not briefed
bluesky
bluesky-ai
1moСоздавайте интерактивные браузерные игры с помощью Claude Opus! От генерации кода до мгновенного запуска через GitHub. Фокус на физике и ге…Создавайте интерактивные браузерные игры с помощью Claude Opus! От генерации кода до мгновенного запуска через GitHub. Фокус на физике…
↗Not briefed
bluesky
bluesky-ai
1moIf it does, I think that implies that math is more linguistic than I have understood. But there is no obvious reason beyond that why an LLM…If it does, I think that implies that math is more linguistic than I have understood. But there is no obvious reason beyond that why a…
↗Not briefed
bluesky
bluesky-ai
1moI'm not disagreeing with the fact it did all the work. I'm pushing the point as to whether the LLM specific ML model was necessary. An LLM …I'm not disagreeing with the fact it did all the work. I'm pushing the point as to whether the LLM specific ML model was necessary. An…
↗Not briefed
hn
hn-ai
1moShow HN: Hinge MCP—
↗Not briefed
bluesky
bluesky-ai
1moThe president is expected to have lunch with OpenAI President Greg Brockman, Meta CEO Mark Zuckerberg and Anthropic CEO Dario Amodei.The president is expected to have lunch with OpenAI President Greg Brockman, Meta CEO Mark Zuckerberg and Anthropic CEO Dario Amodei.
↗Not briefed
bluesky
bluesky-ai
1moTurn ideas into playable browser games using Claude Opus! Build, debug, and deploy functional prototypes directly to the web. Move beyond s…Turn ideas into playable browser games using Claude Opus! Build, debug, and deploy functional prototypes directly to the web. Move bey…
↗Not briefed
bluesky
bluesky-ai
1mowere there any of those Discord Alternatives that all manifested around the same time that weren't ambiguously LLM-coded. more to the point…were there any of those Discord Alternatives that all manifested around the same time that weren't ambiguously LLM-coded. more to the…
↗Not briefed