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
1mo“Accept bad things happening” says guy involved in this case: www.reuters.com/legal/govern...“Accept bad things happening” says guy involved in this case: www.reuters.com/legal/govern...
↗Not briefed
hn
hn-ai
1moThat Thing with Pterodactyl and Blueprint—
↗Not briefed
arxiv
arxiv-ai
1moErrors of LLM-Assisted Literature Retrieval in Environmental Science: A Comparison Study of Abstract versus Full-text Based Prompts🔧 ChatGPTLarge language models (LLMs) are increasingly used for literature search and synthesis. However, it is unclear whether they retrieve a…
↗Not briefed
arxiv
arxiv-ai
1moToward AI Trustworthiness: Finding Analytically Proven Forward-Invariant Sets for AI-Controlled SystemsNeural-network (NN) controllers are increasingly used in nonlinear control systems, but their highly nonlinear behavior makes them dif…
↗Not briefed
arxiv
arxiv-ai
1moSquare-Root Regret for Adversarial Multiplayer Bandits without Collision Information or Shared RandomnessWe study adversarial multiplayer bandits with $K$ arms and $2\le m<K$ labeled players, without collision information, shared randomnes…
↗Not briefed
rss
techmeme
1moChange.org says it is investing $100M of its own money to rebuild its core petitions platform with AI and has launched an AI copilot beta for petition creators (Dan Primack/Axios)Dan Primack / Axios: Change.org says it is investing $100M of its own money to rebuild its core petitions platform with AI and has lau…
↗Not briefed
arxiv
arxiv-ai
1moDo Time-Series QA Systems Read the Time Series? Evidence Use and Reasoning ReliabilityIn recent years, time-series question answering (QA) systems have made significant progress. However, generating a correct answer does…
↗Not briefed
arxiv
arxiv-ai
1moSpend Bytes on Breadth: Precision-Count Trade-offs for Decode-Time KV Compression in Long Chain-of-Thought ReasoningReasoning models write most of their KV cache while decoding long chains of thought (CoT), so the cache has to be compressed online un…
↗Not briefed
arxiv
arxiv-ai
1moTwo-Sample Testing via Path-based InferenceModern deep generative models are primarily studied for their ability to generate realistic samples, yet the generative dynamics they…
↗Not briefed
bluesky
bluesky-ai
1moimo i think this is downstream of how heavily focused we are on social stuff, llm dialect is the uncanny valley for our mental model of the…imo i think this is downstream of how heavily focused we are on social stuff, llm dialect is the uncanny valley for our mental model o…
↗Not briefed
bluesky
bluesky-ai
1moNewton’s husband or boyfriend or whatever works for Anthropic. Tarbell is an AI/EA cult thing. There’s nothing independent going on.Newton’s husband or boyfriend or whatever works for Anthropic. Tarbell is an AI/EA cult thing. There’s nothing independent going on.
↗Not briefed
hn
hn-ai
1moPowerless F1 drivers frustrated by Bahrain F1 software glitch—
↗Not briefed
hn
hn-ai
1moA tribute to one of the best games on the Atari 2600—
↗Not briefed
arxiv
arxiv-ai
1moRESOLVE: Language-Agnostic Validation of GPU Kernels Through Testing, Reduction, and ProofAI systems can now write and optimize production GPU kernels, but validating them remains an important challenge. Evaluating the kerne…
↗Not briefed
arxiv
arxiv-ai
1moKnowing the Rules, Applying the Rules: Evaluating Language Models on Traditional Chinese BaziKnowing domain rules does not guarantee applying them to a case. We study this distinction in traditional Chinese Bazi through 3,000 C…
↗Not briefed
arxiv
arxiv-ai
1moAutomatic Speech Recognition for Low-Resource Sinhala: A Critical Review of Methods, Challenges, and Future Directions🔧 WhisperAutomatic speech recognition (ASR) for low-resource languages remains a major challenge. Sinhala, the primary language of Sri Lanka wi…
↗Not briefed
bluesky
bluesky-ai
1moIf you're going to go with voice actors you hire online, rather than meet in person, do the leg work to vet their work product and ensure i…If you're going to go with voice actors you hire online, rather than meet in person, do the leg work to vet their work product and ens…
↗Not briefed
bluesky
bluesky-ai
1moNEW OpenAI CEO Sam Altman says the world should accept “some bad things happening,” in exchange for the enormous benefits of AI. Via De…NEW OpenAI CEO Sam Altman says the world should accept “some bad things happening,” in exchange for the enormous benefits of AI. V…
↗Not briefed
rss
marktechpost
1moCan an Open Model Do Security Research? Cantina’s apex-flash-1 Solves 40 of 60 Held-Out Bug Tasks🔧 vLLM🔧 Hugging FaceCantina Security, with Yeta Labs, has released apex-flash-1, an open-weights model trained specifically for vulnerability research. It…
↗Not briefed
bluesky
bluesky-ai
1moAltman: “we believe that the world should accept some bad things happening for the benefits of this technology and people having the agency…Altman: “we believe that the world should accept some bad things happening for the benefits of this technology and people having the a…
↗Not briefed
bluesky
bluesky-ai
1moDavid Robinson, who worked on OpenAI’s safety team, resigned from the company this week. “I believe we need to look deeper than specific ru…David Robinson, who worked on OpenAI’s safety team, resigned from the company this week. “I believe we need to look deeper than specif…
↗Not briefed
arxiv
arxiv-ai
1moSharp Integrality Gaps in Calibration DistanceWe study the offline gap between deterministic calibration distance C and its fractional relaxation L for binary unit-weight sequences…
↗Not briefed
bluesky
bluesky-ai
1moI believe Annie Altman and you can too. I’ve had the pleasure of knowing this kind soul, and I hope she is supported and protected by many …I believe Annie Altman and you can too. I’ve had the pleasure of knowing this kind soul, and I hope she is supported and protected by…
↗Not briefed
arxiv
arxiv-ai
1moAn evolutionary origin of collective decision making in humans and machinesGroups of individuals can solve collective problems more accurately than any single member, by aggregating their opinions. Recent theo…
↗Not briefed
bluesky
bluesky-ai
1moAnother possibility is that humans lie in the "low samples, high compute" quadrant that research hasn't really explored. Until recently a b…Another possibility is that humans lie in the "low samples, high compute" quadrant that research hasn't really explored. Until recentl…
↗Not briefed
bluesky
bluesky-ai
1moLawyer up if you want to get paid it’s open season on openaiLawyer up if you want to get paid it’s open season on openai
↗Not briefed
arxiv
arxiv-ai
1moHow Should a Prompt Optimizer Spend a Tight Budget? BudgetAPO with Noise-Adaptive EvaluationAutomatic prompt optimization (APO) has been widely employed to adapt large language models without updating their weights, yielding p…
↗Not briefed
hn
hn-ai
1moShow HN: HyperCMD, write CLI applications using plain HTMLI&#x27;ve been working on this repo, it allows you to write CLI applications using plain HTML. For the behavioral side, you write the…
↗Not briefed
hn
hn-ai
1moL402 edge cache for AI agents paying over Lightning—
↗Not briefed
bluesky
bluesky-ai
1moAnthropic has already signed a deal on its $32bn Dalby data centre before any approvals. Residents haven't been consulted & the Crisafulli …Anthropic has already signed a deal on its $32bn Dalby data centre before any approvals. Residents haven't been consulted & the Crisaf…
↗Not briefed
bluesky
bluesky-ai
1moだからamazonはうちは事業者(法的人格)と契約しているだけで雇用していませんよーってやってるわけです。事業者は言うまでもなく、業務委託・契約者として配達員に主体性(個人契約などの法的人格)を持たせることで、負うべき責任を回避しているんですね。これと同じことをテック企業はLL…だからamazonはうちは事業者(法的人格)と契約しているだけで雇用していませんよーってやってるわけです。事業者は言うまでもなく、業務委託・契約者として配達員に主体性(個人契約などの法的人格)を持たせることで、負うべき責任を回避しているんですね。これと同じことをテック…
↗Not briefed
bluesky
bluesky-ai
1molike, let's grant for a second that underneath the hype there's a technology with genuine applications. the average person still should not…like, let's grant for a second that underneath the hype there's a technology with genuine applications. the average person still shoul…
↗Not briefed
hn
hn-ai
1moWho is cleaning up all the garbage LLMs generate?So my one terabyte drive got full and I realized after months of CLI sessions I generated a lot of markdown files in a lot of images i…
↗Not briefed
bluesky
bluesky-ai
1moalso what higher level ideas ? everything they "come up with" probably came directly from the LLM with little to no second draftsalso what higher level ideas ? everything they "come up with" probably came directly from the LLM with little to no second drafts
↗Not briefed
bluesky
bluesky-ai
1moThere are a myriad of ways of achieving a codebase that, when compiled, results in the same ROM. LLMs will often bruteforce this process an…There are a myriad of ways of achieving a codebase that, when compiled, results in the same ROM. LLMs will often bruteforce this proce…
↗Not briefed
bluesky
bluesky-ai
1moExcept in the LLM case we don’t actually figure it outExcept in the LLM case we don’t actually figure it out
↗Not briefed
hn
hn-ai
1moShow HN: Orrery, a Three.js solar system visualization powered by real NAIF dataHi HN, I&#x27;m excited to share a personal project that I&#x27;ve put a lot of love into.<p>This is Orrery[0], a Three.js-based visua…
↗Not briefed
bluesky
bluesky-ai
1moAnd of course you have cross-over events like when Palantir and Anthropic caused the US to bomb a girls school in Iran.And of course you have cross-over events like when Palantir and Anthropic caused the US to bomb a girls school in Iran.
↗Not briefed
arxiv
arxiv-ai
1moBellman-Centric Learning: Near-Optimal Regret for Linear Bandits with MemoryWe study linear bandits with memory, where past actions induce endogenous nonstationarity through an arbitrary known, bounded matrix-v…
↗Not briefed
bluesky
bluesky-ai
1mo‘Top Chinese models lag their US rivals by just 3% on benchmark scores after the September release of DeepSeek’s V4.1 Flash. That’s down fr…‘Top Chinese models lag their US rivals by just 3% on benchmark scores after the September release of DeepSeek’s V4.1 Flash. That’s do…
↗Not briefed
bluesky
bluesky-ai
1moI don't have a favorite Anthropic whistleblower or a so called one.I don't have a favorite Anthropic whistleblower or a so called one.
↗Not briefed
bluesky
bluesky-ai
1moSam Altman: “the world should accept I’m a shitty person & am going go do shitty things to everyone for money.” Fuck you, Sam Altman, also…Sam Altman: “the world should accept I’m a shitty person & am going go do shitty things to everyone for money.” Fuck you, Sam Altman,…
↗Not briefed
bluesky
bluesky-ai
1moI want the Democrats to run on immuring this man and his employees.I want the Democrats to run on immuring this man and his employees.
↗Not briefed
rss
bing-news-ai
1moTrump unveils 'Super Intelligence Force' to oversee AI policyUS President Donald Trump says he has created a new taskforce focused on artificial intelligence, which will be led by Director of Nat…
↗Not briefed
rss
bing-news-ai
1moSaudi Arabia-based Idraj.AI raises US$598,000 to bring artificial intelligence to the IPO process in the KingdomSaudi Arabia-based artificial intelligence (AI) company Idraj.AI has closed a US$598,000 pre-seed funding round led by ASAP Legal Tech…
↗Not briefed
rss
bing-news-ai
1moSaudi Arabia-based Idraj.AI raises US$598,000 to bring artificial intelligence to the IPO process in the KingdomSaudi Arabia-based artificial intelligence (AI) company Idraj.AI has closed a US$598,000 pre-seed funding round led by ASAP Legal Tech…
↗Not briefed
arxiv
arxiv-ai
1moGraph Data Augmentation via Contrastive Generator Inversion ($\texttt{DCBA}$)Graphs provide a natural representation of many complex systems, ranging from social platforms to ecosystems. However, the development…
↗Not briefed
rss
techmeme
1moQ&A with Google SVP and DeepMind Institute co-director James Manyika on AI risks and why responsibility must be shared across industry, government, and society (Mishal Husain/Bloomberg)Mishal Husain / Bloomberg: Q&A with Google SVP and DeepMind Institute co-director James Manyika on AI risks and why responsibility mus…
↗Not briefed
bluesky
bluesky-ai
1moBudgeting for AI at work is tricky. AI use is measured in tokens, but—as this visual explainer demonstrates—it’s nearly impossible to predi…Budgeting for AI at work is tricky. AI use is measured in tokens, but—as this visual explainer demonstrates—it’s nearly impossible to…
↗Not briefed
rss
bing-news-ai
1moTrump’s latest move on ‘super intelligence’ sparks ridiculeTrump’s latest move on ‘super intelligence’ sparks ridicule ...
↗Not briefed