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
2moI wish I could be excited for this and Layton as well, but given that Level 5 is deep in the ragu when it comes to adopting LLM slop, I jus…I wish I could be excited for this and Layton as well, but given that Level 5 is deep in the ragu when it comes to adopting LLM slop,…
↗Not briefed
hn
hn-ai
2moKnowledge vs. wisdom: asking AI "What mushroom is that?"—
↗Not briefed
bluesky
bluesky-ai
2moUS-Behörden warnen: Chinesische KI-Anbieter wie DeepSeek und Alibaba sollen massiv Wissen aus US-Modellen wie denen von OpenAI oder Google …US-Behörden warnen: Chinesische KI-Anbieter wie DeepSeek und Alibaba sollen massiv Wissen aus US-Modellen wie denen von OpenAI oder Go…
↗Not briefed
hn
hn-ai
2moA language, a runtime and preludes built for LLMs, not for humans—
↗Not briefed
rss
techmeme
2moEnterprise AI security startup Cymphony emerges from stealth with $30M, including a $25M Series A co-led by Sequoia and SMBC, at a $100M+ post-money valuation (Jagmeet Singh/TechCrunch)Jagmeet Singh / TechCrunch: Enterprise AI security startup Cymphony emerges from stealth with $30M, including a $25M Series A co-led b…
↗Not briefed
bluesky
bluesky-ai
2moThe Discovery When a language model thinks aloud, it writes.The Discovery When a language model thinks aloud, it writes.
↗Not briefed
bluesky
bluesky-ai
2moI think that there probably are people at Anthropic who believe that this is all bullshit and want this stuff out there to promote how powe…I think that there probably are people at Anthropic who believe that this is all bullshit and want this stuff out there to promote how…
↗Not briefed
arxiv
arxiv-ai
2moWhat Should an Agent Forget? Separating What Is Stored from What Is UsedPersistent language agents need stored experience to remain available across time, while each answer requires evidence suited to a par…
↗Not briefed
bluesky
bluesky-ai
2moI’m not saying they are bulletproof in their principles, Anthropic has become even more of a profiteer* than OpenAI, but that is what most …I’m not saying they are bulletproof in their principles, Anthropic has become even more of a profiteer* than OpenAI, but that is what…
↗Not briefed
bluesky
bluesky-ai
2moThis AI is an existential threat to humanity The AI:This AI is an existential threat to humanity The AI:
↗Not briefed
bluesky
bluesky-ai
2moBoth Anthropic and OpenAI should face criminal action for felony hacking. We put people in jail for hacking, including people who use scrip…Both Anthropic and OpenAI should face criminal action for felony hacking. We put people in jail for hacking, including people who use…
↗Not briefed
hn
hn-ai
2moPublic vs. Private LLM—
↗Not briefed
bluesky
bluesky-ai
2mobUt ThEy’Re ThE gOoD aI GuYs 🥴bUt ThEy’Re ThE gOoD aI GuYs 🥴
↗Not briefed
bluesky
bluesky-ai
2moFascinating. OpenAI says it has offered proof of Navier-Stokes equations breaking down & creating a singularity in finite time. An exam…Fascinating. OpenAI says it has offered proof of Navier-Stokes equations breaking down & creating a singularity in finite time. An…
↗Not briefed
bluesky
bluesky-ai
2moAnthropic secret advanced model can know simultaneously the exact position and the exact momentum of a quantum particleAnthropic secret advanced model can know simultaneously the exact position and the exact momentum of a quantum particle
↗Not briefed
bluesky
bluesky-ai
2moImo anytime an openai/anthropic employee posts something you should treat it like a commercial. Yes even the latest thingImo anytime an openai/anthropic employee posts something you should treat it like a commercial. Yes even the latest thing
↗Not briefed
bluesky
bluesky-ai
2moI think also because a lot of them are in cults they think the AI will form cults to carry things out which, like, let's be real, it might…I think also because a lot of them are in cults they think the AI will form cults to carry things out which, like, let's be real, it…
↗Not briefed
bluesky
bluesky-ai
2mo"ex-OpenAI employee delivers chilling report that the computer has turned into a higher being that can destroy 75% of civilization by snapp…"ex-OpenAI employee delivers chilling report that the computer has turned into a higher being that can destroy 75% of civilization by…
↗Not briefed
bluesky
bluesky-ai
2moThe genuine answer, of how they rationalize working on it, is that they tell themselves they’re in a race to build the “good” superintellig…The genuine answer, of how they rationalize working on it, is that they tell themselves they’re in a race to build the “good” superint…
↗Not briefed
bluesky
bluesky-ai
2moI quit my job in January at a data company for nonprofits after stating my “ethical, environmental, even spiritual concerns” a day after ne…I quit my job in January at a data company for nonprofits after stating my “ethical, environmental, even spiritual concerns” a day aft…
↗Not briefed
arxiv
arxiv-ai
2moDiSCo: A Distribution-First Steering and Cultural Prior Evaluation Framework for Measuring Cultural Preference Bias in LLMsLarge language models (LLMs) are increasingly deployed in globally used assistants, yet their default choices in culturally grounded e…
↗Not briefed
bluesky
bluesky-ai
2moAs a rule, I mostly hate the “students are cheaters” narrative, but my hate also extends to profs using crappy detector tools and the use o…As a rule, I mostly hate the “students are cheaters” narrative, but my hate also extends to profs using crappy detector tools and the…
↗Not briefed
hn
hn-ai
2moRooting a Sony DPT-RP1 E-Reader for Maximum Manga Enjoyment—
↗Not briefed
bluesky
bluesky-ai
2mo"But China might do this!" is not an answer to "why are American companies doing this?" because this is clearly a situation where OpenAI an…"But China might do this!" is not an answer to "why are American companies doing this?" because this is clearly a situation where Open…
↗Not briefed
bluesky
bluesky-ai
2moThere is really no economic reason to do the LLM cybersecurity stuff. They're hitting the ceiling of LLM code (and I've read people are not…There is really no economic reason to do the LLM cybersecurity stuff. They're hitting the ceiling of LLM code (and I've read people ar…
↗Not briefed
bluesky
bluesky-ai
2moAs I went into with Cal Newport, this is not "autonomous hacking" by "agents that escape sandboxes," it's Anthropic and OpenAI throwing unl…As I went into with Cal Newport, this is not "autonomous hacking" by "agents that escape sandboxes," it's Anthropic and OpenAI throwin…
↗Not briefed
bluesky
bluesky-ai
2moRead the beginning of Aaron's post on AI - the report is frightening, backed up by the warnings from people who left their jobs working on …Read the beginning of Aaron's post on AI - the report is frightening, backed up by the warnings from people who left their jobs workin…
↗Not briefed
bluesky
bluesky-ai
2moso there's a small cottage industry of "AI has gained SuperGod Status and will wipe out half of humanity" stories lately and my very simple…so there's a small cottage industry of "AI has gained SuperGod Status and will wipe out half of humanity" stories lately and my very s…
↗Not briefed
bluesky
bluesky-ai
2moAlso, we LONG AGO passed the point where AI can in principle cause a nuclear war by accident. Automated warning systems are primitive late…🔧 ChatGPTAlso, we LONG AGO passed the point where AI can in principle cause a nuclear war by accident. Automated warning systems are primitive…
↗Not briefed
hn
hn-ai
2moGPT-6 Astra, Looped Transformers, and Hidden Reasoning—
↗Not briefed
bluesky
bluesky-ai
2moScoop: OpenAI CEO Sam Altman privately told a group of conservatives that he did not support having the government take shares in the compa…Scoop: OpenAI CEO Sam Altman privately told a group of conservatives that he did not support having the government take shares in the…
↗Not briefed
arxiv
arxiv-ai
2moA-JIT: Agentic Just-In-Time Software ConstructionTraditional software delivery assumes a static paradigm: code is constructed prior to execution and deployed as a fixed artifact. We p…
↗Not briefed
bluesky
bluesky-ai
2mo“This podcast is sponsored by Anthropic. Anthropic: the AI company that thinks there’s a greater than 10% chance their product will kill al…“This podcast is sponsored by Anthropic. Anthropic: the AI company that thinks there’s a greater than 10% chance their product will ki…
↗Not briefed
bluesky
bluesky-ai
2morecall how last winter the big Anthropic discount promo that got a ton of people moving on Claude Code was during December, when people are…recall how last winter the big Anthropic discount promo that got a ton of people moving on Claude Code was during December, when peopl…
↗Not briefed
bluesky
bluesky-ai
2moAI has a 10% chance of blowing up the world, according to the maverick new LLM, Roko's Bullshitter.AI has a 10% chance of blowing up the world, according to the maverick new LLM, Roko's Bullshitter.
↗Not briefed
hn
hn-ai
2moNew results in square packing problem by a hobbyist—
↗Not briefed
rss
techmeme
2moSources: Dell seeks to raise ~$4B via an investment-grade bond sale to refinance outstanding notes due in 2026, amid the AI-driven demand for servers (Bloomberg)Bloomberg: Sources: Dell seeks to raise ~$4B via an investment-grade bond sale to refinance outstanding notes due in 2026, amid the AI…
↗Not briefed
arxiv
arxiv-ai
2moTwo-Token Features and Small-Large Ensembles for VLM Hallucination DetectionWe present our system for the SHROOM-Visions 2026 shared task on character-level VLM hallucination detection. A small ($4$B-parameter)…
↗Not briefed
bluesky
bluesky-ai
2moThree Anthropic researchers went public last night with chilling concerns about out-of-control AI, warning it could destroy humans this dec…Three Anthropic researchers went public last night with chilling concerns about out-of-control AI, warning it could destroy humans thi…
↗Not briefed
bluesky
bluesky-ai
2moOpenAI CEO Sam Altman privately told a group of conservatives this summer that he did not support having the government take shares in the …OpenAI CEO Sam Altman privately told a group of conservatives this summer that he did not support having the government take shares in…
↗Not briefed
bluesky
bluesky-ai
2mo6 years ago when i first experimented with LLM technology and learned about the speculative extreme risks, i was not convinced the technolo…6 years ago when i first experimented with LLM technology and learned about the speculative extreme risks, i was not convinced the tec…
↗Not briefed
arxiv
arxiv-ai
2moLiteRAG: Cost-Efficient Graph-Based Retrieval-Augmented GenerationGraph-based retrieval can improve multi-hop question answering, but existing approaches often incur high query-time costs and produce…
↗Not briefed
bluesky
bluesky-ai
2moSo my understanding is that both of the proposed N-S solutions involved AI, the second being the researchers allegedly plagiarized by OpenA…So my understanding is that both of the proposed N-S solutions involved AI, the second being the researchers allegedly plagiarized by…
↗Not briefed
bluesky
bluesky-ai
2moprivacy.openai.comprivacy.openai.com
↗Not briefed
hn
hn-ai
2moPythagorean Lambdoma Harmonic Keyboard Created by Barbara Hero—
↗Not briefed
hn
hn-ai
2moX402-trinity – kills the cloud server back end for local AI and robotics—
↗Not briefed
bluesky
bluesky-ai
2moTo be clear, there is an actual threat from LLMs: Anthropic and OpenAI using hundreds of billions of dollars' worth of infrastructure provi…To be clear, there is an actual threat from LLMs: Anthropic and OpenAI using hundreds of billions of dollars' worth of infrastructure…
↗Not briefed
arxiv
arxiv-ai
2moThe Answer Path and the Grounding Instruction in LLM Question Answering over Knowledge GraphsA graph retrieval-augmented generation pipeline chooses which triples to put in the prompt, a syntax to write them in, an order to wri…
↗Not briefed
hn
hn-ai
2moShow HN: Emailclaw–Email lovers' AI agent that creates scheduled task in 5 steps—
↗Not briefed
bluesky
bluesky-ai
2moWhát I dó is me: for that I came (an LLM could never)Whát I dó is me: for that I came (an LLM could never)
↗Not briefed