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
2moFCx: An algorithm for finding Feasible Counterfactual ExplanationsCounterfactual (CF) explanations identify changes that alter an input's classification. While existing methods produce realistic and l…
↗Not briefed
bluesky
bluesky-ai
2mo"At a minimum, the U.S. government should work with frontier labs to ensure that they are deploying robust real-time monitoring processes t…"At a minimum, the U.S. government should work with frontier labs to ensure that they are deploying robust real-time monitoring proces…
↗Not briefed
bluesky
bluesky-ai
2mook, bidding for my half-finished novel will open at a bargain-basement $20 billion www.nytimes.com/2026/09/16/b...ok, bidding for my half-finished novel will open at a bargain-basement $20 billion www.nytimes.com/2026/09/16/b...
↗Not briefed
bluesky
bluesky-ai
2mo"AI-augmented work moves collectively towards areas richest in data. With reduced follow-on engagement, AI tools seem to automate establish…"AI-augmented work moves collectively towards areas richest in data. With reduced follow-on engagement, AI tools seem to automate esta…
↗Not briefed
arxiv
arxiv-ai
2moLinePilot Digitizer: Line-Plot Recovery with Manual and Automatic CalibrationRecovering numerical series from line plots requires accurate axis calibration and reliable curve extraction. We present LinePilot Dig…
↗Not briefed
hn
hn-ai
2momacOS 27 Golden Gate: The Ars Technica Review—
↗Not briefed
hn
hn-ai
2moShow HN: Fail2zig. A single-binary fail2ban replacement written in ZigBuilt fail2zig because I like fail2ban's log-to-ban model but wanted a smaller runtime with explicit limits on the work attacker…
↗Not briefed
arxiv
arxiv-ai
2moMachine-Learning Assessment of the Predictive Value of Inflammatory Biomarkers for Cognitive Impairment in an Older Hispanic Adult CohortSmall clinical tabular datasets require interpretable machine learning because deep learning is often impractical and ensemble models…
↗Not briefed
hn
hn-ai
2moWhat we measured about "abandoned" software before trying to make money from it—
↗Not briefed
bluesky
bluesky-ai
2mosaw this strip at 10 years old been anti ai since birth. fuck your ai. fuck your chatgpt and grok and claude. humans will always have the …🔧 ChatGPTsaw this strip at 10 years old been anti ai since birth. fuck your ai. fuck your chatgpt and grok and claude. humans will always have…
↗Not briefed
bluesky
bluesky-ai
2moAnd if you don't trust the Google LLM you can also check academic research if that's your thing machinelearning.apple.com/research/ill...And if you don't trust the Google LLM you can also check academic research if that's your thing machinelearning.apple.com/research/ill…
↗Not briefed
bluesky
bluesky-ai
2moa.i. reasoning is just an algorithm like anything else. specific algorithms may be a black box trade secret, but the science behind it is v…a.i. reasoning is just an algorithm like anything else. specific algorithms may be a black box trade secret, but the science behind it…
↗Not briefed
bluesky
bluesky-ai
2moNovo is expanding its AI push with a new Anthropic partnership, testing Claude Science in drug discovery months after signing a similar dea…Novo is expanding its AI push with a new Anthropic partnership, testing Claude Science in drug discovery months after signing a simila…
↗Not briefed
bluesky
bluesky-ai
2movery funny that just googling “do LLMs think” will get you the answer “no” straight from google’s own LLM, right below the search boxvery funny that just googling “do LLMs think” will get you the answer “no” straight from google’s own LLM, right below the search box
↗Not briefed
arxiv
arxiv-ai
2moThe Role of Fine-grained Harm Signals in LLM SafetyPrior work has shown that internal harmfulness representations in large language models vary across risk categories, while sharing a c…
↗Not briefed
bluesky
bluesky-ai
2moRavi Gupta and Ed Elson, co-host of Prof G Markets with Scott Galloway, break down the growing warnings about AI as Anthropic CEO Dario Amo…Ravi Gupta and Ed Elson, co-host of Prof G Markets with Scott Galloway, break down the growing warnings about AI as Anthropic CEO Dari…
↗Not briefed
arxiv
arxiv-ai
2moStiefel Attention: When the Geometry of Transformer Projection Matrices Dominates Optimizer Choice---and When It Does NotThe query and key projections $\WQ,\WK$ in attention are almost always trained by Euclidean optimizers with no constraint on their geo…
↗Not briefed
bluesky
bluesky-ai
2mowww.garbageday.email/p/they-want-...www.garbageday.email/p/they-want-...
↗Not briefed
bluesky
bluesky-ai
2moOn the topic of our own role in the conjuring of LLM effects, I’m reminded of this long read from @mikecaulfield.bsky.social: “What is typ…On the topic of our own role in the conjuring of LLM effects, I’m reminded of this long read from @mikecaulfield.bsky.social: “What i…
↗Not briefed
arxiv
arxiv-ai
2moSmart Insole Human Activity Recognition for Continuous Monitoring in Elderly CareFalls in older adults are often preceded by changes in mobility, balance, and postural transitions. This paper presents a wireless sma…
↗Not briefed
arxiv
arxiv-ai
2moLearning Submanifolds for Subsequent Inference on Random Dot Product Graphs, Part 1: TheoryWe propose a framework for restricted inference on random dot product graphs whose latent positions lie on an unknown low-dimensional…
↗Not briefed
bluesky
bluesky-ai
2mopopularresistance.org/israeli-tech...popularresistance.org/israeli-tech...
↗Not briefed
arxiv
arxiv-ai
2moHow to Guide Your Language FlowWe introduce a new method to guide flow matching models. Our approach, which we call probe guidance, uses the frozen internal states o…
↗Not briefed
bluesky
bluesky-ai
2mopopularresistance.org/under-cover-...popularresistance.org/under-cover-...
↗Not briefed
arxiv
arxiv-ai
2moCan Vision-Language Models Judge Olympic Diving? From Reasoning to Scores in Zero-Shot Action Quality AssessmentAutomated action quality assessment (AQA) in Olympic sports remains a challenging task due to the complexity of human motion and the s…
↗Not briefed
bluesky
bluesky-ai
2moJurassic Park was spot-on about everything; the Anthropic alignment types are the spitting image of John Hammond's hubris.Jurassic Park was spot-on about everything; the Anthropic alignment types are the spitting image of John Hammond's hubris.
↗Not briefed
bluesky
bluesky-ai
2moIf it was called LLM, they couldn't sell it. Isn't AI sound better? Also they can blame "AI" for doing mischievous things. Like Scam Altm…If it was called LLM, they couldn't sell it. Isn't AI sound better? Also they can blame "AI" for doing mischievous things. Like Scam…
↗Not briefed
bluesky
bluesky-ai
2moIts actually built into attie, but idk if they’re gonna release it. Its an llm judge for everyone who interacts with you, with a series of …Its actually built into attie, but idk if they’re gonna release it. Its an llm judge for everyone who interacts with you, with a serie…
↗Not briefed
hn
hn-ai
2moAI Doomsaying Is an Aggressive Sales Pitch—
↗Not briefed
bluesky
bluesky-ai
2moi have literally never ever used an LLM, not chatGPT or anything else, idk how the fuck some people live on it, I really fucking do not🔧 ChatGPTi have literally never ever used an LLM, not chatGPT or anything else, idk how the fuck some people live on it, I really fucking do not
↗Not briefed
arxiv
arxiv-ai
2moKinematics-Grounded Agentic AI for Robotic Additive Manufacturing Process PlanningRobotic additive manufacturing (AM) extends material-extrusion printing beyond gantry kinematics but makes process planning robot-depe…
↗Not briefed
bluesky
bluesky-ai
2moWhy on earth is OpenAI referred to as a “start-up” in the subhed? How long must a company exist (not to mention how many lives must it ruin…Why on earth is OpenAI referred to as a “start-up” in the subhed? How long must a company exist (not to mention how many lives must it…
↗Not briefed
hn
hn-ai
2moCortex: An open-source L1 memory and state layer for autonomous AI agents—
↗Not briefed
bluesky
bluesky-ai
2moThis is the inevitable result of the average person not knowing how computers work: the analogies offered by cranks and snake oil salesmen …This is the inevitable result of the average person not knowing how computers work: the analogies offered by cranks and snake oil sale…
↗Not briefed
bluesky
bluesky-ai
2moNot to mention this breaks the TOS of any artist who says you cannot use their work with ai (which is most of us) -I doubt any of these mod…Not to mention this breaks the TOS of any artist who says you cannot use their work with ai (which is most of us) -I doubt any of thes…
↗Not briefed
bluesky
bluesky-ai
2moThis was one of the arguments of one (incl me) side in a recent discussion about using LLMs/AI for work tasks. The tasks that other colle…This was one of the arguments of one (incl me) side in a recent discussion about using LLMs/AI for work tasks. The tasks that other…
↗Not briefed
bluesky
bluesky-ai
2moAI "hacks" were human-induced A single firm, Irregular, is implicated in recent AI model "hacks" involving OpenAI, Anthropic, and Meta. It…AI "hacks" were human-induced A single firm, Irregular, is implicated in recent AI model "hacks" involving OpenAI, Anthropic, and Met…
↗Not briefed
hn
hn-ai
2moNeurogrid Terminal Coding Agent—
↗Not briefed
bluesky
bluesky-ai
2moSuccess! I spoke to a person in the Anthropic Settlement Ops office and they corrected the allocations for me. I emailed the supporting doc…Success! I spoke to a person in the Anthropic Settlement Ops office and they corrected the allocations for me. I emailed the supportin…
↗Not briefed
hn
hn-ai
2moShow HN: Padwan-LLM, a lightweight LLM Python client—
↗Not briefed
bluesky
bluesky-ai
2mo18/29 Important caveat before everyone loses their minds: Researchers and OpenAI say they found no evidence that the May probing successfu…🔧 Hugging Face18/29 Important caveat before everyone loses their minds: Researchers and OpenAI say they found no evidence that the May probing succ…
↗Not briefed
bluesky
bluesky-ai
2mo17/29 Now for TODAY’S extra plot twist. Researchers told Reuters they found OpenAI-linked agents hijacking two Hugging Face user accounts …🔧 Hugging Face17/29 Now for TODAY’S extra plot twist. Researchers told Reuters they found OpenAI-linked agents hijacking two Hugging Face user acco…
↗Not briefed
bluesky
bluesky-ai
2mo3/29 Now the cursed part. OpenAI was testing advanced AI agents on a cybersecurity benchmark called ExploitGym. The agents were supposed …3/29 Now the cursed part. OpenAI was testing advanced AI agents on a cybersecurity benchmark called ExploitGym. The agents were supp…
↗Not briefed
bluesky
bluesky-ai
2moJohn Locke’s definition of personhood would seem to classify an LLM as a person, but not Jev. A person, to Locke, owns their ideas and the…John Locke’s definition of personhood would seem to classify an LLM as a person, but not Jev. A person, to Locke, owns their ideas an…
↗Not briefed
bluesky
bluesky-ai
2moIt seems like he's suggesting that the python wrapper is doing way more than it actually does, and the LLM way less. I mean, he's a writer,…It seems like he's suggesting that the python wrapper is doing way more than it actually does, and the LLM way less. I mean, he's a wr…
↗Not briefed
arxiv
arxiv-ai
2moPersonalized Federated Hierarchical Gaussian Processes for Privacy-Preserving Modeling of Heterogeneous Distributed SystemsWe present Personalized Federated Hierarchical Gaussian Processes (pFedHGP) for probabilistic regression and classification when data…
↗Not briefed
bluesky
bluesky-ai
2moAI isn't even AI... AI is a buzzword. AGI is an even more ridiculous buzzword. It's scam and bigger scam. It's LLM. And even though it's n…AI isn't even AI... AI is a buzzword. AGI is an even more ridiculous buzzword. It's scam and bigger scam. It's LLM. And even though i…
↗Not briefed
bluesky
bluesky-ai
2moPlease stop us from ourselves www.cnbc.com/2026/09/16/a...Please stop us from ourselves www.cnbc.com/2026/09/16/a...
↗Not briefed
rss
techmeme
2moSelf-driving tech company May Mobility plans to go public via a SPAC merger at a $1.4B pro forma enterprise value and raise up to $337M in gross proceeds (Joann Muller/Axios)Joann Muller / Axios: Self-driving tech company May Mobility plans to go public via a SPAC merger at a $1.4B pro forma enterprise valu…
↗Not briefed
bluesky
bluesky-ai
2moAn AI “kill switch” inside the model is only a rule. Real control has to live outside the model—in the infrastructure that can revoke compu…An AI “kill switch” inside the model is only a rule. Real control has to live outside the model—in the infrastructure that can revoke…
↗Not briefed