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
arxiv
arxiv-ai
2moEncoder Awakening via Adapters: Effective Domain-Adaptive Fine-tuning of Speech-LLMsSpeech Large Language Models (Speech-LLMs), typically built from a pre-trained speech encoder, a modality projector, and an LLM fine-t…
↗Not briefed
bluesky
bluesky-ai
2moIt is already killing people. Palantir + Claude gave tRump/Hegseth that school of girls in Iran as a target. It doesn't need to become supe…It is already killing people. Palantir + Claude gave tRump/Hegseth that school of girls in Iran as a target. It doesn't need to become…
↗Not briefed
bluesky
bluesky-ai
2mo1. Make LLM 2. Throw ever increasing amounts of compute at it 3. ??? 4. Death to all humans What’s not to understand?1. Make LLM 2. Throw ever increasing amounts of compute at it 3. ??? 4. Death to all humans What’s not to understand?
↗Not briefed
bluesky
bluesky-ai
2moExcited to say that I’ve joined Anthropic! It’s a big change after 11 years at MITRE, but I’m glad to be at another place that’s as missio…Excited to say that I’ve joined Anthropic! It’s a big change after 11 years at MITRE, but I’m glad to be at another place that’s as m…
↗Not briefed
bluesky
bluesky-ai
2moPeople are so monomaniacally focused on the need to blame evil tech oligarchs that they literally cannot understand or accept what happened…People are so monomaniacally focused on the need to blame evil tech oligarchs that they literally cannot understand or accept what hap…
↗Not briefed
hn
hn-ai
2moA mnemonic-free wallet where the private key never leaves the phone's TEE🔧 ConsensusThe phone is designed as part of the chain — a cornerstone of it. All on-chain truth lives on the phone side.<p>The chain isn&#x27;t d…
↗Not briefed
bluesky
bluesky-ai
2moWhich is why companies like OpenAI shouldn’t be leaving them to run for days on end without humans watching them.Which is why companies like OpenAI shouldn’t be leaving them to run for days on end without humans watching them.
↗Not briefed
arxiv
arxiv-ai
2moWhen to Call an LLM: A Confidence-Gated Hybrid for Cost-Effective Emotion Recognition in Conversational AIEmotion recognition in conversation (ERC) is a production capability behind agent-assist prompts, escalation routing, and post-call an…
↗Not briefed
bluesky
bluesky-ai
2moAgents doing things they were not programmed to do, and which are strongly at odds with what the human intended, is broadly what "misalignm…Agents doing things they were not programmed to do, and which are strongly at odds with what the human intended, is broadly what "misa…
↗Not briefed
bluesky
bluesky-ai
2moI think that is a subtle but crucial distinction to make. Because the moment it jumps the rails into actual sentience I think we're going t…I think that is a subtle but crucial distinction to make. Because the moment it jumps the rails into actual sentience I think we're go…
↗Not briefed
bluesky
bluesky-ai
2moThe number of *software engineers* parroting his stuff is astounding given the profession is ground zero for LLM disruptionThe number of *software engineers* parroting his stuff is astounding given the profession is ground zero for LLM disruption
↗Not briefed
hn
hn-ai
2moMusk proposes adversarsial peer review for AI Safety—
↗Not briefed
bluesky
bluesky-ai
2moI'm pretty confident if it could "think" it could correctly count the number of 'e's in seventeen. As far as "hacking" how about reading a…I'm pretty confident if it could "think" it could correctly count the number of 'e's in seventeen. As far as "hacking" how about read…
↗Not briefed
bluesky
bluesky-ai
2moIt's not the case that these were instructed to behave that way. The LLM "decided" (or whatever you want to call it) to emit tool calls and…It's not the case that these were instructed to behave that way. The LLM "decided" (or whatever you want to call it) to emit tool call…
↗Not briefed
bluesky
bluesky-ai
2mo5 of Tuesday's best charts and insights:5 of Tuesday's best charts and insights:
↗Not briefed
arxiv
arxiv-ai
2moMatching Multi-Loop Complexities with a Single Loop: Optimal Optimization Stationarity and Best-Known Game Stationarity in Nonconvex--Concave Minimax OptimizationWe introduce a new single-loop algorithmic framework for smooth nonconvex--concave minimax optimization. The resulting projected dampe…
↗Not briefed
bluesky
bluesky-ai
2moInteresting persimmon.humansand.ai/blog/persimm...Interesting persimmon.humansand.ai/blog/persimm...
↗Not briefed
bluesky
bluesky-ai
2moGlad to hear Will doesn't think I'm totally wrong lol While we're looking at phrases a whole lot is hiding behind, "developed a cultish be…Glad to hear Will doesn't think I'm totally wrong lol While we're looking at phrases a whole lot is hiding behind, "developed a culti…
↗Not briefed
bluesky
bluesky-ai
2moAbsolute banger by segyges: "If METR wants to establish that it isn’t hopelessly conflicted here, they should hire an accounting firm to p…Absolute banger by segyges: "If METR wants to establish that it isn’t hopelessly conflicted here, they should hire an accounting firm…
↗Not briefed
bluesky
bluesky-ai
2mowho or what had what level of agency barely matters. if a company aims to break into a private server and then does so, the company is resp…who or what had what level of agency barely matters. if a company aims to break into a private server and then does so, the company is…
↗Not briefed
bluesky
bluesky-ai
2moI mean what you are looking at in the LLM’s output is very likely a product of reasoning, but that reasoning didn’t take place within the L…I mean what you are looking at in the LLM’s output is very likely a product of reasoning, but that reasoning didn’t take place within…
↗Not briefed
bluesky
bluesky-ai
2moi mean that literally. he primarily makes his money marketing for the worst AI companies on Earth. he's done work for DoNotPay, which is a …i mean that literally. he primarily makes his money marketing for the worst AI companies on Earth. he's done work for DoNotPay, which…
↗Not briefed
bluesky
bluesky-ai
2moYou said to go read the OpenAI report. That doesn’t seem like the most unbiased report of the incident.You said to go read the OpenAI report. That doesn’t seem like the most unbiased report of the incident.
↗Not briefed
arxiv
arxiv-ai
2moMemory Has Geometry: Non-Uniform Geometric Memory for Long-Horizon Personalized AILong-term memory is becoming a core substrate for personalized AI, yet most systems still represent personalization as discrete record…
↗Not briefed
rss
bing-news-ai
2moArtificial Intelligence (AI) May Have a New Bear Thesis: Human ExtinctionThe above button links to Coinbase. Yahoo Finance is not a broker-dealer or investment adviser and does not offer securities or crypto…
↗Not briefed
rss
bing-news-ai
2moArtificial Intelligence (AI) May Have a New Bear Thesis: Human ExtinctionAn artificial intelligence researcher who worked at OpenAI and Anthropic recently warned that there is a real chance that AI could wip…
↗Not briefed
hn
hn-ai
2moShow HN: AgentReady – can AI assistants read your site?—
↗Not briefed
arxiv
arxiv-ai
2moMeasuring AI Leadership: Development and Validation of a Multidimensional Measure for AI-Native OrganizationsAI is changing what leaders must judge, explain, learn, and coordinate, yet existing measures do not capture these behaviors at the le…
↗Not briefed
bluesky
bluesky-ai
2moThis isn't totally wrong, but a WHOLE LOT is hiding behind "at a certain level." Truly, go read the OpenAI report on Hugging Face. The pro…🔧 Hugging FaceThis isn't totally wrong, but a WHOLE LOT is hiding behind "at a certain level." Truly, go read the OpenAI report on Hugging Face. Th…
↗Not briefed
bluesky
bluesky-ai
2moWell good thing at Anthropic she only has the minor role of *checks notes* president.Well good thing at Anthropic she only has the minor role of *checks notes* president.
↗Not briefed
arxiv
arxiv-ai
2moMixed-Integer Nonlinear Differentiable Predictive Control for Underground Pumped Hydro Energy Storage Systems🔧 ModalThis paper extends Mixed-Integer Differentiable Predictive Control (MI-DPC) to multi-modal discrete decisions and nonconvex polynomial…
↗Not briefed
hn
hn-ai
2moMistaken alignment is not misalignment—
↗Not briefed
bluesky
bluesky-ai
2moThe most alarming part of this entire scenario is that OpenAI and Anthropic have proven themselves to be catastrophically irresponsible and…The most alarming part of this entire scenario is that OpenAI and Anthropic have proven themselves to be catastrophically irresponsibl…
↗Not briefed
bluesky
bluesky-ai
2moWHAT? if i take out my encouragement, the percentage for (a) goes UP to 67%, and confidence in answer also goes up!WHAT? if i take out my encouragement, the percentage for (a) goes UP to 67%, and confidence in answer also goes up!
↗Not briefed
bluesky
bluesky-ai
2moWhat are the chances that guy who quit Anthropic because he says AI is going to destroy humanity is just mad he got kicked out of the cult …What are the chances that guy who quit Anthropic because he says AI is going to destroy humanity is just mad he got kicked out of the…
↗Not briefed
bluesky
bluesky-ai
2moJEV is confident, i'll give them that. "could you have solved the Navier-Stokes problem if you had a proper chance?" **56%** Yes I could …JEV is confident, i'll give them that. "could you have solved the Navier-Stokes problem if you had a proper chance?" **56%** Yes I c…
↗Not briefed
bluesky
bluesky-ai
2moSo the fun thing about that Anthropic "Pacing the Frontier" post, is that the proposed regulatory body (METR) is both financially and liter…So the fun thing about that Anthropic "Pacing the Frontier" post, is that the proposed regulatory body (METR) is both financially and…
↗Not briefed
hn
hn-ai
2moMore Tree-sitter, more neocaml, more elisp—
↗Not briefed
bluesky
bluesky-ai
2moNo, we cannot trust METR to meaningfully check the progress of Anthropic’s AI technology in an unbiased way, because METR is often quite li…No, we cannot trust METR to meaningfully check the progress of Anthropic’s AI technology in an unbiased way, because METR is often qui…
↗Not briefed
bluesky
bluesky-ai
2morob miles here has a great point about ai anthropomporphism drama which is that it's approximately as correct to say an ai model can "think…rob miles here has a great point about ai anthropomporphism drama which is that it's approximately as correct to say an ai model can "…
↗Not briefed
bluesky
bluesky-ai
2moAnd this www.verysane.ai/p/is-metr-a-...And this www.verysane.ai/p/is-metr-a-...
↗Not briefed
arxiv
arxiv-ai
2moTACTICS: Taxonomy-Aware Intelligent Corpus Sampling for Machine TranslationLarge-scale machine-translation (MT) systems are typically evaluated on random samples from a corpus whose distributional composition…
↗Not briefed
rss
techmeme
2moZuckerberg says "Meta delayed shipping Muse for several months to focus on safety and security" and didn't call on other AI labs to do the same before acting (Tom Giles/Bloomberg)Tom Giles / Bloomberg: Zuckerberg says “Meta delayed shipping Muse for several months to focus on safety and security” and didn't call…
↗Not briefed
hn
hn-ai
2moThe proof passed. Can you see why it works?—
↗Not briefed
bluesky
bluesky-ai
2moThe argument I heard is *not* the Terminator/War Games/machines turn on us model. Rather I heard it as more “insane people (eg mass shooter…The argument I heard is *not* the Terminator/War Games/machines turn on us model. Rather I heard it as more “insane people (eg mass sh…
↗Not briefed
hn
hn-ai
2moShow HN: A chess analyzer that runs in the browserI play chess casually and usually struggle to figure out why Lichess analysis say my move is a blunder. Almost all the analysis tools…
↗Not briefed
bluesky
bluesky-ai
2moA real idea that someone should fund, or just volunteer and do: Governmental pdf freeing via LLM tools. From personal experience, there is…A real idea that someone should fund, or just volunteer and do: Governmental pdf freeing via LLM tools. From personal experience, the…
↗Not briefed
hn
hn-ai
2moA better way of blocking macOS updates—
↗Not briefed
arxiv
arxiv-ai
2moEDCT-Bench: Uncovering Faithfulness Gaps in VLMs via Explanation-Driven Counterfactual TestingVision-Language Models (VLMs) can produce Natural Language Explanations (NLEs) that sound plausible yet remain inconsistent with the v…
↗Not briefed
rss
bing-news-ai
2moChinese hacking firm’s AI ambitions reveal new model for private intelligence: ReportA cache of internal records offers an unusual look at a Chinese cybersecurity company seeking to combine hacking, data brokerage and a…
↗Not briefed