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
1moOpenAI shared this as part of a disclosure that one of their agents tried to escape its sandbox to access the internet to answer a question…OpenAI shared this as part of a disclosure that one of their agents tried to escape its sandbox to access the internet to answer a que…
↗Not briefed
bluesky
bluesky-ai
1moOpenAI just shared that all training, evaluation, and tool usage by its most capable models is paused. In a world where social media compa…OpenAI just shared that all training, evaluation, and tool usage by its most capable models is paused. In a world where social media…
↗Not briefed
hn
hn-ai
1moJuhRadial MX: Linux control for MX devices and generic—
↗Not briefed
bluesky
bluesky-ai
1moI'm going to become the joker about how every video essay from a channel I don't recognise I've clicked on this week has been blatantly gen…I'm going to become the joker about how every video essay from a channel I don't recognise I've clicked on this week has been blatantl…
↗Not briefed
hn
hn-ai
1moOpenAI says its AI agents posted user images online in error—
↗Not briefed
bluesky
bluesky-ai
1moGood. Would you prefer ignorant and stupid AI? @richove.bsky.social Oxford lets OpenAI train its AI models on Bodleian Library www.theguar…Good. Would you prefer ignorant and stupid AI? @richove.bsky.social Oxford lets OpenAI train its AI models on Bodleian Library www.th…
↗Not briefed
bluesky
bluesky-ai
1mopeople shit on openai & anthropic for a lot of things, but imo they’ve done a lot to push AI into the discourse and have society decide for…people shit on openai & anthropic for a lot of things, but imo they’ve done a lot to push AI into the discourse and have society decid…
↗Not briefed
bluesky
bluesky-ai
1moYep. Some good analysis over on Substack concerning the shit show that is about to unfold. Oracle has future booked revenue of 300b from Op…Yep. Some good analysis over on Substack concerning the shit show that is about to unfold. Oracle has future booked revenue of 300b fr…
↗Not briefed
bluesky
bluesky-ai
1moOpenAI agents posted user images online, disclose dozens of third party incidents www.axios.com/2026/09/25/o...OpenAI agents posted user images online, disclose dozens of third party incidents www.axios.com/2026/09/25/o...
↗Not briefed
bluesky
bluesky-ai
1moJournalists have fallen in love with the expression "went rogue." It's not appropriate. OpenAI is recklessly releasing malware: software in…Journalists have fallen in love with the expression "went rogue." It's not appropriate. OpenAI is recklessly releasing malware: softwa…
↗Not briefed
bluesky
bluesky-ai
1moCorrect headline: "OpenAI attacks US Government by hacking website." The word "rogue" is all about avoiding accountability, and there is z…Correct headline: "OpenAI attacks US Government by hacking website." The word "rogue" is all about avoiding accountability, and there…
↗Not briefed
bluesky
bluesky-ai
1moThat one is definitely an LLM thoughThat one is definitely an LLM though
↗Not briefed
bluesky
bluesky-ai
1mowww.bbc.com/news/article...www.bbc.com/news/article...
↗Not briefed
bluesky
bluesky-ai
1moI do not use Google as search any more because of the AI imposition. It's been about a year and a half now. I also don't use Adobe products…I do not use Google as search any more because of the AI imposition. It's been about a year and a half now. I also don't use Adobe pro…
↗Not briefed
hn
hn-ai
1moShow HN: Ollaya Web UI—
↗Not briefed
bluesky
bluesky-ai
1moOxford lets OpenAI train its AI models on Bodleian Library University staff voice concerns over reputational risk of partnering with compa…🔧 ChatGPTOxford lets OpenAI train its AI models on Bodleian Library University staff voice concerns over reputational risk of partnering with…
↗Not briefed
bluesky
bluesky-ai
1moEdward Snowden still lives in exile and has a 30-year jail sentence hanging over his head for a security breach 1% as serious as this Sinc…Edward Snowden still lives in exile and has a 30-year jail sentence hanging over his head for a security breach 1% as serious as this…
↗Not briefed
hn
hn-ai
1moClaude Opus 5.5 Should Raise Your Ambitions—
↗Not briefed
bluesky
bluesky-ai
1mothe cool thing about social media in the year of our lord twenty hundred & twenty six is it's impossible to tell whether you're being scold…the cool thing about social media in the year of our lord twenty hundred & twenty six is it's impossible to tell whether you're being…
↗Not briefed
bluesky
bluesky-ai
1monos.nl/l/2632578 ai is niet veilig, ai is moeilijk te controleren, laat staan onder controle te houden. Dit toont het nog maar eens aan.nos.nl/l/2632578 ai is niet veilig, ai is moeilijk te controleren, laat staan onder controle te houden. Dit toont het nog maar eens aa…
↗Not briefed
bluesky
bluesky-ai
1moThis you?This you?
↗Not briefed
bluesky
bluesky-ai
1mo#AI #ChatGPT #OpenAI 🌐 www.newsmart.ch🔧 ChatGPT#AI #ChatGPT #OpenAI 🌐 www.newsmart.ch
↗Not briefed
bluesky
bluesky-ai
1moits really remarkable about how all of the vocabulary in this cultist-speak music video originated before LLMs and has little to do with th…its really remarkable about how all of the vocabulary in this cultist-speak music video originated before LLMs and has little to do wi…
↗Not briefed
bluesky
bluesky-ai
1moOpenAIs KI platzierte Bilder seiner Nutzer auf Online-Plattformen. Es ist der erste bekannte Sicherheitsvorfall bei dem ChatGPT-Nutzerdaten…🔧 ChatGPTOpenAIs KI platzierte Bilder seiner Nutzer auf Online-Plattformen. Es ist der erste bekannte Sicherheitsvorfall bei dem ChatGPT-Nutzer…
↗Not briefed
bluesky
bluesky-ai
1moYou are still on KLM? Most of the world has already moved to LLM.You are still on KLM? Most of the world has already moved to LLM.
↗Not briefed
bluesky
bluesky-ai
1moFuturistic Model #QP TFTT @nevyn79.bsky.social #Aiart #Aiイラスト #aiart #Art #A.IArt #aiartcommunity #midjourney #art #synthart #DigitalA…Futuristic Model #QP TFTT @nevyn79.bsky.social #Aiart #Aiイラスト #aiart #Art #A.IArt #aiartcommunity #midjourney #art #synthart #Dig…
↗Not briefed
bluesky
bluesky-ai
1moFuturistic Model TFTT @nevyn79.bsky.social #Aiart #Aiイラスト #aiart #Art #A.IArt #aiartcommunity #midjourney #art #synthart #DigitalArt #G…Futuristic Model TFTT @nevyn79.bsky.social #Aiart #Aiイラスト #aiart #Art #A.IArt #aiartcommunity #midjourney #art #synthart #DigitalA…
↗Not briefed
bluesky
bluesky-ai
1moSecond PodCastComic, a deepdive on the smart-stupid debate I started. Does AI have to be smart to be deadly? uh...NO! The AI-features list…Second PodCastComic, a deepdive on the smart-stupid debate I started. Does AI have to be smart to be deadly? uh...NO! The AI-features…
↗Not briefed
hn
hn-ai
1moShow HN: I built a tool that gives any website an API and MCP🔧 ChatGPTAnyone else run into this problem?<p>I built staaake Meets (staaakemeets.com) because we run side tech events around the world. You cr…
↗Not briefed
hn
hn-ai
1moAn API that beats GPT-4V, Gemini and Claude at detecting photo rotation—
↗Not briefed
bluesky
bluesky-ai
1mo好きゲームの新作も新グッズも素直に喜べないの本当にシンドイわね…… 純粋に好きな人たちからはアンチ判定だしな……(どんだけ🧱されていることか……) まぁある意味アンチですLLM使うのやめろ好きゲームの新作も新グッズも素直に喜べないの本当にシンドイわね…… 純粋に好きな人たちからはアンチ判定だしな……(どんだけ🧱されていることか……) まぁある意味アンチですLLM使うのやめろ
↗Not briefed
bluesky
bluesky-ai
1moNo. Just no. A catastrophically stupid decision, no doubt fuelled by money. www.theguardian.com/technology/2...No. Just no. A catastrophically stupid decision, no doubt fuelled by money. www.theguardian.com/technology/2...
↗Not briefed
bluesky
bluesky-ai
1moAI agents from OpenAI infiltrated U.S. government websites due to a glitch during testing. The agents created fake email addresses, bypass…AI agents from OpenAI infiltrated U.S. government websites due to a glitch during testing. The agents created fake email addresses, b…
↗Not briefed
hn
hn-ai
1moBook Review: Lee Kuan Yew's Memoirs—
↗Not briefed
bluesky
bluesky-ai
1moAs I struggle more and more with the question of whether a story written by AI can still be a good story, I find this piece by Mike Monteir…As I struggle more and more with the question of whether a story written by AI can still be a good story, I find this piece by Mike Mo…
↗Not briefed
bluesky
bluesky-ai
1mo🙀 LLM Policies: Progress At All Costs diegoe.be/2026/09/25/l... #AI #opensource🙀 LLM Policies: Progress At All Costs diegoe.be/2026/09/25/l... #AI #opensource
↗Not briefed
bluesky
bluesky-ai
1moOxford lets OpenAI train its AI models on Bodleian libraryOxford lets OpenAI train its AI models on Bodleian library
↗Not briefed
bluesky
bluesky-ai
1moA powerful board at OpenAI, the safety and security committee, is facing scrutiny from government as rogue AI agent disclosures continue w…A powerful board at OpenAI, the safety and security committee, is facing scrutiny from government as rogue AI agent disclosures contin…
↗Not briefed
bluesky
bluesky-ai
1moOracle is trying to delay payments due to circumstances beyond their control, however due to the amount of borrowing Oracle has done to inv…Oracle is trying to delay payments due to circumstances beyond their control, however due to the amount of borrowing Oracle has done t…
↗Not briefed
hn
hn-ai
1moLearned Helplessness and Medium's Stats—
↗Not briefed
bluesky
bluesky-ai
1moApple SiliconがローカルLLM使うのにコスパいいのね。いまさら理解。Apple SiliconがローカルLLM使うのにコスパいいのね。いまさら理解。
↗Not briefed
bluesky
bluesky-ai
1moOpenAI said governments could be among "dozens" of entities around the world that may be infiltrated by its artificial intelligence models,…OpenAI said governments could be among "dozens" of entities around the world that may be infiltrated by its artificial intelligence mo…
↗Not briefed
bluesky
bluesky-ai
1moAs OpenAI disclosed "dozens" more breaches, new evidence on Australia's incident reveals that the company's agents spent a week persistentl…As OpenAI disclosed "dozens" more breaches, new evidence on Australia's incident reveals that the company's agents spent a week persis…
↗Not briefed
hn
hn-ai
1moOpenAI Says Its Models Engaged with US Government Websites in New Disclosure—
↗Not briefed
bluesky
bluesky-ai
1moOpenAI has revealed six cases of what it calls "model misalignment" in its experimental AI systems. The most alarming: an unreleased resear…OpenAI has revealed six cases of what it calls "model misalignment" in its experimental AI systems. The most alarming: an unreleased r…
↗Not briefed
bluesky
bluesky-ai
1moOpenAI models accessed US government data and 'tried to hack education site'OpenAI models accessed US government data and 'tried to hack education site'
↗Not briefed
bluesky
bluesky-ai
1moC'est pas vraiment un mystère puisque j'en ai déjà parlé ( notamment sur discord et instagram ) ; mais je bosse actuellement sur une update…C'est pas vraiment un mystère puisque j'en ai déjà parlé ( notamment sur discord et instagram ) ; mais je bosse actuellement sur une u…
↗Not briefed
bluesky
bluesky-ai
1moOpen AI has informed dozens of parties that their websites may have been impacted by their models! This is unacceptable and should result…Open AI has informed dozens of parties that their websites may have been impacted by their models! This is unacceptable and should r…
↗Not briefed
bluesky
bluesky-ai
1moHackers hijack AI accounts and servers to fuel new cyber crime boom ft.trib.al/kuYFEIKHackers hijack AI accounts and servers to fuel new cyber crime boom ft.trib.al/kuYFEIK
↗Not briefed
hn
hn-ai
1moAres: a cross-platform, open-source, multi-system emulator—
↗Not briefed