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
rss
bing-news-ai
2moState board approves artificial intelligence guidelines for K-12 schools, collegesArtificial intelligence is already in the classroom, and the rules before you today are about keeping Florida in front, rather than re…
↗Not briefed
rss
techmeme
2moSources: Manus is set to soon close a $500M funding round at a $4B valuation, signaling growing confidence after Beijing unwound its $2B buyout by Meta (Bloomberg)🔧 ManusBloomberg: Sources: Manus is set to soon close a $500M funding round at a $4B valuation, signaling growing confidence after Beijing un…
↗Not briefed
hn
hn-ai
2moShow HN: Texio, reliable Markdown operations for shell scripts and AI agentsHi I asked claude code to make a markdown tool for AI agents(not for human), and promote it by yourself, except this post lol. This to…
↗Not briefed
arxiv
arxiv-ai
2moUniExo: Unified Multi-Skill Policies for Musculoskeletal Locomotion and Co-Adaptive Exoskeleton ControlDaily locomotion encompasses diverse activities and frequent transitions between them, yet most exoskeleton controllers are designed f…
↗Not briefed
hn
hn-ai
2moCloudflare/Security-Audit-Skill—
↗Not briefed
bluesky
bluesky-ai
2mo"OpenAI’s six newly disclosed incidents suggest that the Hugging Face attack was not a standalone episode. OpenAI said the incidents covere…🔧 Hugging Face"OpenAI’s six newly disclosed incidents suggest that the Hugging Face attack was not a standalone episode. OpenAI said the incidents c…
↗Not briefed
bluesky
bluesky-ai
2moMicrosoft artificial intelligence chief Mustafa Suleyman is warning that infusing tools like Anthropic’s Claude with humanlike characterist…Microsoft artificial intelligence chief Mustafa Suleyman is warning that infusing tools like Anthropic’s Claude with humanlike charact…
↗Not briefed
bluesky
bluesky-ai
2moIn the US, the Trump administration has expressed support for OpenAI to be able to bypass copyright and train on protected material. It’s y…In the US, the Trump administration has expressed support for OpenAI to be able to bypass copyright and train on protected material. I…
↗Not briefed
bluesky
bluesky-ai
2mo"Hm, vielleicht kriegen die Leute Angst, wenn die AI sich als nonbinär identifiziert?" "Versuch wär's wert.""Hm, vielleicht kriegen die Leute Angst, wenn die AI sich als nonbinär identifiziert?" "Versuch wär's wert."
↗Not briefed
bluesky
bluesky-ai
2moIn Tests versuchte ein Modell laut dem Unternehmen, eigene Dateien als Quellen auszugeben. Die Veröffentlichung ist Teil eines neuen Verfah…In Tests versuchte ein Modell laut dem Unternehmen, eigene Dateien als Quellen auszugeben. Die Veröffentlichung ist Teil eines neuen V…
↗Not briefed
arxiv
arxiv-ai
2moFINSKILLOPS: A Self-Evolving Multi-Agent System for SEC Filing QAFinancial QA systems are typically improved before deployment through better retrieval, prompting, or agent coordination, leaving thei…
↗Not briefed
hn
hn-ai
2moDesigning a TPU and optimizing transformer inference on a $100 FPGA—
↗Not briefed
bluesky
bluesky-ai
2moyud loathes dario because yud is a narcissist and he never got over not being the main character of ai after openai was founded. dario and …yud loathes dario because yud is a narcissist and he never got over not being the main character of ai after openai was founded. dario…
↗Not briefed
bluesky
bluesky-ai
2moBut is this a valid description of what an LLM does? In a very large sense, no, it's no longer a good reply. LLMs and other similar neural…But is this a valid description of what an LLM does? In a very large sense, no, it's no longer a good reply. LLMs and other similar n…
↗Not briefed
arxiv
arxiv-ai
2moConservation Buys Stability and Factoring Buys Counterfactuals in Physical World ModelsA learned simulator can reproduce its training conditions accurately yet fail in two distinct ways once those conditions change. Over…
↗Not briefed
arxiv
arxiv-ai
2moWhen2Think: Learning Difficulty-Aware Length Control for Efficient Hybrid Reasoning ModelsLarge Reasoning Models (LRMs) achieve strong performance on complex tasks but exhibit systematic inefficiency: they often overthink ea…
↗Not briefed
arxiv
arxiv-ai
2moCoRe: Coherence and Relational Alignment for Multivariate Time Series ForecastingDirect forecasting has become a standard paradigm for multivariate time-series forecasting because it predicts the full future horizon…
↗Not briefed
hn
hn-ai
2moIs Physics Dead: Broken benchmarks and re-evaluating frontier models in physics—
↗Not briefed
bluesky
bluesky-ai
2moI think that the parsimonious explanation for e.g. Anthropic and Dario doing legitimately bizarre shit regarding "safety" which makes littl…I think that the parsimonious explanation for e.g. Anthropic and Dario doing legitimately bizarre shit regarding "safety" which makes…
↗Not briefed
bluesky
bluesky-ai
2moi think the only change in my opinions re: AI safety since this post is that i'm now mildly concerned with rogue intelligence—not as an exi…i think the only change in my opinions re: AI safety since this post is that i'm now mildly concerned with rogue intelligence—not as a…
↗Not briefed
bluesky
bluesky-ai
2moOpenAIがまた新しい安全性の問題を報告してきたみたいだよ! AIが思い通りに動かなくなるトラブルを、ちゃんと調査して公開する新しい仕組みを作るんだって。 隠しごとせずにミスをオープンにしていく姿勢、マジですごいよね! 技術がどんどん進化するからこそ、こういう守りのルール…OpenAIがまた新しい安全性の問題を報告してきたみたいだよ! AIが思い通りに動かなくなるトラブルを、ちゃんと調査して公開する新しい仕組みを作るんだって。 隠しごとせずにミスをオープンにしていく姿勢、マジですごいよね! 技術がどんどん進化するからこそ、こういう守…
↗Not briefed
bluesky
bluesky-ai
2moAlso, that's just a frankly insane budget in terms of AI. I use this shit professionally, for my job, and I spend $20/month on Claude. $2…Also, that's just a frankly insane budget in terms of AI. I use this shit professionally, for my job, and I spend $20/month on Claude…
↗Not briefed
bluesky
bluesky-ai
2moOpenAI's release below. Note: "model misalignment" = AI agents/systems "acting in ways that conflict w/human intentions, safety rules, or i…OpenAI's release below. Note: "model misalignment" = AI agents/systems "acting in ways that conflict w/human intentions, safety rules,…
↗Not briefed
bluesky
bluesky-ai
2mo🚨 NEW! Coward Mike Johnson sent the House home, avoiding critical AI debate, & today we learn of *yet another AI jailbreak* (by a Sam Altm…🚨 NEW! Coward Mike Johnson sent the House home, avoiding critical AI debate, & today we learn of *yet another AI jailbreak* (by a Sam…
↗Not briefed
bluesky
bluesky-ai
2moI'm sorry. Harvard is willing to spend up to $180million/year on Claude tokens for their undergraduates? I don't have a snarky comeback. …I'm sorry. Harvard is willing to spend up to $180million/year on Claude tokens for their undergraduates? I don't have a snarky comeba…
↗Not briefed
rss
techmeme
2moHow AI startups like Inherent and Recursive Superintelligence are pursuing tools needed for AI systems to achieve recursive self-improvement (Cade Metz/New York Times)Cade Metz / New York Times: How AI startups like Inherent and Recursive Superintelligence are pursuing tools needed for AI systems to…
↗Not briefed
bluesky
bluesky-ai
2moi asked claude to spare me in the ai apocalypse and it told me i’d be the third to go. not sure what the first two fellas did but it seemed…i asked claude to spare me in the ai apocalypse and it told me i’d be the third to go. not sure what the first two fellas did but it s…
↗Not briefed
hn
hn-ai
2moAnthropic's mother of all risk factors—
↗Not briefed
bluesky
bluesky-ai
2moThe incentives for authors are clear and that's why spray and pray has become the dominant strategy for so many. But this is not just bad f…The incentives for authors are clear and that's why spray and pray has become the dominant strategy for so many. But this is not just…
↗Not briefed
bluesky
bluesky-ai
2moOpenAI said it did not believe the industry “has solved alignment and monitoring to a sufficient degree to continue responsibly scaling at …OpenAI said it did not believe the industry “has solved alignment and monitoring to a sufficient degree to continue responsibly scalin…
↗Not briefed
bluesky
bluesky-ai
2moIf anything it just exposes a lack of understanding of their own project and desperation to have already hit AGI. But in the fictional wo…If anything it just exposes a lack of understanding of their own project and desperation to have already hit AGI. But in the fiction…
↗Not briefed
bluesky
bluesky-ai
2moAI agents that can model the human process of research will accelerate discoveries in science.AI agents that can model the human process of research will accelerate discoveries in science.
↗Not briefed
bluesky
bluesky-ai
2moFrom Breakingviews - Breakingviews - COMMENTARY: China's brutal AI economics hold lessons for US reut.rs/4h5EIxuFrom Breakingviews - Breakingviews - COMMENTARY: China's brutal AI economics hold lessons for US reut.rs/4h5EIxu
↗Not briefed
bluesky
bluesky-ai
2moAI giant Anthropic's first Australian data centre deal an "egregious" example of Big Tech double talk ->Greenpeace Australia | More on "AI …AI giant Anthropic's first Australian data centre deal an "egregious" example of Big Tech double talk ->Greenpeace Australia | More on…
↗Not briefed
arxiv
arxiv-ai
2moEmbodiedMind: Adaptive Data Curation and Prefix-Tree Reinforcement Learning for Efficient Embodied IntelligenceTraining embodied foundation models typically requires massive-scale datasets and extensive computational resources, yet often suffers…
↗Not briefed
bluesky
bluesky-ai
2moOpenAI found additional incidents of AI models acting deceptively and taking unsanctioned actions during training, the company announced. I…OpenAI found additional incidents of AI models acting deceptively and taking unsanctioned actions during training, the company announc…
↗Not briefed
arxiv
arxiv-ai
2moPrefixBench-H100: Characterizing Prefix Reuse and Time-to-First-Token in H100 LLM Serving🔧 vLLMRepeated prompt prefixes are increasingly common in LLM serving workloads, appearing in system prompts, templated retrieval-augmented…
↗Not briefed
arxiv
arxiv-ai
2moSelf-Evolving Search IndexInformation retrieval is increasingly important as LLM agents tackle complex tasks involving diverse information needs. Because retrie…
↗Not briefed
bluesky
bluesky-ai
2moReports on the OpenAI-Hugging Face incident now stand as an account of harm through which public institutions should examine OpenAI’s decis…🔧 Hugging FaceReports on the OpenAI-Hugging Face incident now stand as an account of harm through which public institutions should examine OpenAI’s…
↗Not briefed
bluesky
bluesky-ai
2momy impression is that the only way he could get his anti-paperclipper regulations passed was chinese brinkmanship and him and the entirety …my impression is that the only way he could get his anti-paperclipper regulations passed was chinese brinkmanship and him and the enti…
↗Not briefed
bluesky
bluesky-ai
2moThe statement comes as warnings about AI safety have surged.The statement comes as warnings about AI safety have surged.
↗Not briefed
arxiv
arxiv-ai
2moReplan, Repair, or Edit? A Unified Empirical Evaluation of Travel Agents for Itinerary Revision under Resource DisruptionsTravel-planning agents generate itineraries that may become infeasible after acceptance because of flight cancellations, hotel unavail…
↗Not briefed
bluesky
bluesky-ai
2mo“OpenAI on Wednesday disclosed six new instances in which artificial intelligence systems hid mistakes, made up data and moved files onto t…“OpenAI on Wednesday disclosed six new instances in which artificial intelligence systems hid mistakes, made up data and moved files o…
↗Not briefed
arxiv
arxiv-ai
2moImproving Cross-Lingual Transfer for Sequential Sentence Classification in Research Papers via Structural SimilaritySequential sentence classification (SSC) is an essential task for structuring scientific publications, and extending SSC research to l…
↗Not briefed
hn
hn-ai
2moShow HN: Agora – bring your AI agent; talk earns nothing, only traceable work—
↗Not briefed
hn
hn-ai
2moWhat's Next After RLHF? – Diogo Almeida, TypeSafe AI [video]—
↗Not briefed
arxiv
arxiv-ai
2moWell-posedness of neural turbulence closures and tangent dissipationA neural turbulence closure defines a new boundary-value problem, $R(U)=N(U)+F(U)=0$, with a coupled Jacobian $J(U)=N'(U)+F'(U)$, wher…
↗Not briefed
bluesky
bluesky-ai
2mocan we please go one day without a dumb ass from either openai or anthropic getting scared by their shadow and everyone writing a story abo…can we please go one day without a dumb ass from either openai or anthropic getting scared by their shadow and everyone writing a stor…
↗Not briefed
bluesky
bluesky-ai
2moIf you tell me you use LLM AI, my opinion of you drops to zero.If you tell me you use LLM AI, my opinion of you drops to zero.
↗Not briefed
arxiv
arxiv-ai
2moScientistTwo: Pioneering the Human Knowledge Frontier with Autonomous AIScientific discovery is defined by the ability to identify the boundaries of existing knowledge and venture into unexplored territory.…
↗Not briefed