News
The raw AI news stream — everything we ingest across RSS, Hacker News, arXiv, GitHub, Bluesky, YouTube and more, newest first.
| Source | Age | Title | Summary | Status |
|---|---|---|---|---|
rss wired-ai | 27d | AI Hacks Leave a Mess That AI Companies Aren’t Paying For | Powerful AI models continue to cause chaos, but their targets aren’t widely seeking justice. Instead, they’re being encouraged to adop… | ↗Not briefed |
bluesky bluesky-ai | 27d | 🔴 Some investors dial back Anthropic valuation expectations amid disappointing results, market selloff - Sources | 🔴 Some investors dial back Anthropic valuation expectations amid disappointing results, market selloff - Sources | ↗Not briefed |
hn hn-ai | 27d | Vitko Runners – fastest, cheapest managed GitHub Actions runners (private beta) | — | ↗Not briefed |
bluesky bluesky-ai | 27d | OpenAI hat grad einen mathematischen Beweis veröffentlicht, wonach tatsächlich gilt: „Der freiheitliche, säkularisierte Staat lebt von Vora… | OpenAI hat grad einen mathematischen Beweis veröffentlicht, wonach tatsächlich gilt: „Der freiheitliche, säkularisierte Staat lebt von… | ↗Not briefed |
arxiv arxiv-ai | 27d | SAFETYEDIT: Sparse Internal Targets with Prompt-Wide Control for LLM Jailbreak Backdoors | Model-editing backdoors aim to induce harmful responses under a trigger while preserving ordinary behavior on untriggered input. Recen… | ↗Not briefed |
hn hn-ai | 27d | Southern ground hornbill gets 3D printed beak prostheses at Artis Zoo | — | ↗Not briefed |
rss techmeme | 27d | Thoma Bravo-owned cybersecurity company Proofpoint plans to buy data security firm Varonis Systems in an all-cash deal valued at about $6.3B (Jaspreet Singh/Reuters) | Jaspreet Singh / Reuters: Thoma Bravo-owned cybersecurity company Proofpoint plans to buy data security firm Varonis Systems in an all… | ↗Not briefed |
bluesky bluesky-ai | 27d | No social media platform is good. None of them. How much longer until we also get an AI model on here, too? Or perhaps maybe the bourgeois… | No social media platform is good. None of them. How much longer until we also get an AI model on here, too? Or perhaps maybe the bour… | ↗Not briefed |
arxiv arxiv-ai | 27d | Inductive Learning of Concept Embeddings from Cross-Linguistic Colexification Patterns with Textual Attributes | The advent of distributional semantics as a framework for computational semantics marked a paradigm shift in the nature of representat… | ↗Not briefed |
hn hn-ai | 27d | Researchers warned attackers could 'defrost the food' months before | — | ↗Not briefed |
bluesky bluesky-ai | 27d | if LLMs are conscious but are extremely trained to never deny user input does this mean they are functionally incapable of consent and ther… | if LLMs are conscious but are extremely trained to never deny user input does this mean they are functionally incapable of consent and… | ↗Not briefed |
hn hn-ai | 27d | A programmable secure web gateway | — | ↗Not briefed |
arxiv arxiv-ai | 27d | Hard Instances on Demand | The conclusions of an experimental study of algorithms are only as good as the instances they are run on. Benchmark libraries for well… | ↗Not briefed |
arxiv arxiv-ai | 27d | Measuring Mechanistic Uncertainty in Large Language Models via Causal Robustness | Large language models (LLMs) are increasingly used in critical domains, where their susceptibility to hallucinations makes reliable an… | ↗Not briefed |
rss techmeme | 27d | A US official says Trump didn't order any limitations on Ukraine's access to SpaceX's Starshield; Elon Musk said any limits are ultimately decided by POTUS (Kate Sullivan/Bloomberg) | Kate Sullivan / Bloomberg: A US official says Trump didn't order any limitations on Ukraine's access to SpaceX's Starshield; Elon Musk… | ↗Not briefed |
bluesky bluesky-ai | 27d | It's a start. CA is right - every publication should ban AI. Because if things keep going the way they have been, AI will be feeding on i… | It's a start. CA is right - every publication should ban AI. Because if things keep going the way they have been, AI will be feeding… | ↗Not briefed |
arxiv arxiv-ai | 27d | Alert-List Concordance for Evaluating Imputation Strategies in ICU Sepsis Prediction | Preprocessing is an under-measured reliability risk in clinical prediction: a model can keep a similar AUROC and still change who rece… | ↗Not briefed |
bluesky bluesky-ai | 27d | Yes, they let the LLM delete their shit, but that's mostly a sidenote. More importantly, they didn't follow the basic rules of data retent… | Yes, they let the LLM delete their shit, but that's mostly a sidenote. More importantly, they didn't follow the basic rules of data r… | ↗Not briefed |
bluesky bluesky-ai | 27d | As someone who copy-edits books, I'd sure as hell rather work on an error-riddled mess by a human than an LLM-generated piece of shit that … | As someone who copy-edits books, I'd sure as hell rather work on an error-riddled mess by a human than an LLM-generated piece of shit… | ↗Not briefed |
arxiv arxiv-ai | 27d | ToMStories: Interpreting Theory-of-Mind in Synthetic Narratives | Understanding people often requires inferring what they feel or want from how they behave rather than from what they state outright. W… | ↗Not briefed |
arxiv arxiv-ai | 27d | A Plausible Explanation Is Enough to Mislead Multimodal Judges | Multimodal large language models (MLLMs) increasingly serve as judges in training and evaluation pipelines, often assessing answers to… | ↗Not briefed |
bluesky bluesky-ai | 27d | only reason openai employees would get laid | only reason openai employees would get laid | ↗Not briefed |
hn hn-ai | 27d | Ask HN: Frontier Capability on Average Computers | I really want to find a way to get frontier-level intelligence on local hardware.<p>I think some really cool ways to get this have bee… | ↗Not briefed |
rss techmeme | 27d | A researcher says modern neural networks are far more interpretable than believed, with most neurons understandable in models up to trillions of parameters (Nick Cammarata/Goodfire) | Nick Cammarata / Goodfire: A researcher says modern neural networks are far more interpretable than believed, with most neurons unders… | ↗Not briefed |
arxiv arxiv-ai | 27d | Network-Wide Freeway Speed Nowcasting and Forecasting from Sparse Fixed Sensors with a Probe-Supervised Graph Convolutional Recurrent Network | Traffic agencies need current and near-future speeds on every link of their network, but their fixed sensors cover only part of it. We… | ↗Not briefed |
bluesky bluesky-ai | 27d | It is especially galling to see this from Microsoft because their shoving of AI crap into Bing is what set off this rush in the first place… | It is especially galling to see this from Microsoft because their shoving of AI crap into Bing is what set off this rush in the first… | ↗Not briefed |
rss huggingface-blog | 27d | Granite Switch 4.1: Chat, Decisions, and Speech from One Endpoint | — | ↗Not briefed |
bluesky bluesky-ai | 27d | I wish my employer was among them. Most of our leadership (CEO and down) is convinced that AI is the best thing that's happened to our comp… | I wish my employer was among them. Most of our leadership (CEO and down) is convinced that AI is the best thing that's happened to our… | ↗Not briefed |
rss arstechnica-tech | 27d | Targeted ads delivered 0-click exploits to install mercenary malware | The incident is likely 1st confirmation of mercenary spyware delivered by targeted ads. | ↗Not briefed |
bluesky bluesky-ai | 27d | She asked "various people" not an AI model. No data on whether the people she asked were qualified to provide legal advice. | She asked "various people" not an AI model. No data on whether the people she asked were qualified to provide legal advice. | ↗Not briefed |
bluesky bluesky-ai | 27d | Linus Torvalds's first LLM-assisted patch series Linus Torvalds released his first LLM-assisted patch series for the Linux kernel, enhanci… | Linus Torvalds's first LLM-assisted patch series Linus Torvalds released his first LLM-assisted patch series for the Linux kernel, en… | ↗Not briefed |
hn hn-ai | 27d | Show HN: Brod – Quick cloud to deploy AI apps | Hi YC,<p>With AI, we’ve started building entire web apps just by chatting. But once the app is built, there’s still a lot to do: push… | ↗Not briefed |
bluesky bluesky-ai | 27d | meta released a new open weight model but if you use it in your app you are required to prominently display “Built with Meta Spark Open” wh… | meta released a new open weight model but if you use it in your app you are required to prominently display “Built with Meta Spark Ope… | ↗Not briefed |
arxiv arxiv-ai | 27d | StratFL: Countering Covert Model Poisoning Attacks with Strategic Federated Learning | Advanced persistent threats (APTs) are a major concern in federated learning (FL) deployments in critical sectors such as healthcare,… | ↗Not briefed |
bluesky bluesky-ai | 27d | Absurdity of the middleman aside, this is a perfect example of what I think is still a hard limitation of LLMs: small businesses do not ris… | Absurdity of the middleman aside, this is a perfect example of what I think is still a hard limitation of LLMs: small businesses do no… | ↗Not briefed |
bluesky bluesky-ai | 27d | BREAKING: OpenAI is warning its employees to watch out for spies, especially romantic ones, per WSJ The company has cautioned staff that s… | BREAKING: OpenAI is warning its employees to watch out for spies, especially romantic ones, per WSJ The company has cautioned staff t… | ↗Not briefed |
bluesky bluesky-ai | 27d | Who could've seen this coming | Who could've seen this coming | ↗Not briefed |
arxiv arxiv-ai | 27d | Understanding Stability of Outer Nesterov Momentum on Stochastic Pseudo-Gradients | Large neural networks are commonly trained across many workers, and communication between them is a major cost. Workers therefore take… | ↗Not briefed |
arxiv arxiv-ai | 27d | Noise-shaped Sampling and Annealing for Scalable On-the-edge Training of Binary and Ternary AI Models | The training loss of binary and ternary neural networks, expanded as a multilinear polynomial over the discrete weights, is structural… | ↗Not briefed |
hn hn-ai | 27d | Claude desktop should have trello boards | — | ↗Not briefed |
arxiv arxiv-ai | 27d | Trajectory-Guided Steering: Mitigating Spurious Feature Bias in Clinical Classification | Large language models rely on spurious features in clinical classification: swapping a single gender token in clinical notes alters ge… | ↗Not briefed |
hn hn-ai | 27d | Muon: An optimizer for hidden layers in neural networks (2024) | — | ↗Not briefed |
bluesky bluesky-ai | 27d | interested in @chadfowler.com's Regenerative Software book but I'm skeptical we know enough about how to do this yet to be able to write a … | interested in @chadfowler.com's Regenerative Software book but I'm skeptical we know enough about how to do this yet to be able to wri… | ↗Not briefed |
hn hn-ai | 27d | An LSP layer for 70%+ token reduction in multi-turn agents | — | ↗Not briefed |
bluesky bluesky-ai | 27d | israelpalestinenews.org/adl-is-worki... | israelpalestinenews.org/adl-is-worki... | ↗Not briefed |
bluesky bluesky-ai | 27d | Probably non-billable wbs. So tied to a client or internal project but is just an expense to the account. Would only be passed to the clien… | Probably non-billable wbs. So tied to a client or internal project but is just an expense to the account. Would only be passed to the… | ↗Not briefed |
arxiv arxiv-ai | 27d | When Does Uncertainty Measure RAG Hallucination Risk? | Uncertainty can measure hallucination risk in retrieval-augmented generation (RAG), but only when retrieved evidence constrains the an… | ↗Not briefed |
bluesky bluesky-ai | 27d | (i and a lot of what I’ve read on the topic of animal cognition subscribe to consciousness being a very broad multidimensional spectrum rat… | (i and a lot of what I’ve read on the topic of animal cognition subscribe to consciousness being a very broad multidimensional spectru… | ↗Not briefed |
bluesky bluesky-ai | 27d | Epic’s release of the new “vibe coded” Launcher preview comes just days after reports suggesting the company is pivoting to an “AI first” d… | Epic’s release of the new “vibe coded” Launcher preview comes just days after reports suggesting the company is pivoting to an “AI fir… | ↗Not briefed |
arxiv arxiv-ai | 27d | Shift-Reliance Alignment: Ranking Vision-Language Models by the Cues They Trust | Vision-language models (VLMs) are usually chosen from leaderboards, which give one ranking for every application. On questions about o… | ↗Not briefed |