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 |
|---|---|---|---|---|
arxiv arxiv-ai | 2mo | Logit Refiner: Improving Visual Autoregressive Models via Intra-Scale Dependency Modeling | Visual Autoregressive Models (VAR) generate images through next-scale prediction, producing all tokens within each scale in parallel.… | ↗Not briefed |
bluesky bluesky-ai | 2mo | im not sure what the point of having an academic math community would be in such a scenario. it would genuinely make more sense to just han… | im not sure what the point of having an academic math community would be in such a scenario. it would genuinely make more sense to jus… | ↗Not briefed |
bluesky bluesky-ai | 2mo | Anybody else notice that the dude that tweeted that “all the big AI companies think that AI will kill us all in 3 years”… 1) only worked a… | Anybody else notice that the dude that tweeted that “all the big AI companies think that AI will kill us all in 3 years”… 1) only wor… | ↗Not briefed |
hn hn-ai | 2mo | The Age-Gating of History | — | ↗Not briefed |
bluesky bluesky-ai | 2mo | I'm so sick of these guys. Every journalist who hears them say "AI could kill us all" should turn the question around and say "Why should O… | I'm so sick of these guys. Every journalist who hears them say "AI could kill us all" should turn the question around and say "Why sho… | ↗Not briefed |
arxiv arxiv-ai | 2mo | Thinking with Looped Flows | Humans and machines often solve harder problems by spending more time on computation. In deep learning, looped models implement this i… | ↗Not briefed |
hn hn-ai | 2mo | Show HN: Faint Light, 10-100x faster plate-solving/astrometry in Rust | — | ↗Not briefed |
arxiv arxiv-ai | 2mo | SpecGuard: Inference-Time Backdoor Detection For Free | Large language models are often fine-tuned, shared, or downloaded from third parties, so a deployed model may carry a hidden backdoor… | ↗Not briefed |
hn hn-ai | 2mo | Gitwal – GitHub Event Archive | — | ↗Not briefed |
bluesky bluesky-ai | 2mo | Recently obtained contracts between Big Tech companies & the Pentagon reveal their efforts to game the risk assessment process in AI weapon… | Recently obtained contracts between Big Tech companies & the Pentagon reveal their efforts to game the risk assessment process in AI w… | ↗Not briefed |
hn hn-ai | 2mo | We wanted to use Baseten for inference, we got admin access to their GitHub🔧 Baseten | — | ↗Not briefed |
bluesky bluesky-ai | 2mo | the messaging actually going out is that mathematicians will have value *because* they will be needed to interact with LLMs. perhaps LLM co… | the messaging actually going out is that mathematicians will have value *because* they will be needed to interact with LLMs. perhaps L… | ↗Not briefed |
bluesky bluesky-ai | 2mo | A US government adviser warned of risks to the AI industry amid spreading public and political concern that super-advanced AIs could one da… | A US government adviser warned of risks to the AI industry amid spreading public and political concern that super-advanced AIs could o… | ↗Not briefed |
hn hn-ai | 2mo | Show HN: I built an AI Agent that sell itself | Quick bit of context: SendRoq is a LinkedIn automation tool.<p>Before the AI era (pre-2021), I ran a marketing agency. Back then, clie… | ↗Not briefed |
bluesky bluesky-ai | 2mo | i think Jones and Acemoglu are closer to the market than Anthropic, ngl | i think Jones and Acemoglu are closer to the market than Anthropic, ngl | ↗Not briefed |
bluesky bluesky-ai | 2mo | The good news is with this latest disclosure from Anthropic we finally know about all of the automated cyberattacks that have been carried … | The good news is with this latest disclosure from Anthropic we finally know about all of the automated cyberattacks that have been car… | ↗Not briefed |
bluesky bluesky-ai | 2mo | The LLM is always there, will always respond in a matter of seconds, and is programmed to give output that hits the emotional need centers.… | The LLM is always there, will always respond in a matter of seconds, and is programmed to give output that hits the emotional need cen… | ↗Not briefed |
bluesky bluesky-ai | 2mo | NVIDIA, Google, Apple, Meta, OpenAI, Anthropic all shuttling money around in a big circle jerk, natch. | NVIDIA, Google, Apple, Meta, OpenAI, Anthropic all shuttling money around in a big circle jerk, natch. | ↗Not briefed |
hn hn-ai | 2mo | Gemini Desktop | — | ↗Not briefed |
arxiv arxiv-ai | 2mo | Dynamic language model representations for multi-objective reaction optimisation | Optimising chemical reactions across multiple objectives, such as yield, selectivity, and safety, is central to chemical synthesis, an… | ↗Not briefed |
bluesky bluesky-ai | 2mo | So fucking tired of it. Don't worry, I will still hear about it incessantly on other platforms! | So fucking tired of it. Don't worry, I will still hear about it incessantly on other platforms! | ↗Not briefed |
bluesky bluesky-ai | 2mo | one recurring theme in the coverage of this AI Anthropic doomsday newscycle is nobody actually has any REAL ideas of what to do just a lot… | one recurring theme in the coverage of this AI Anthropic doomsday newscycle is nobody actually has any REAL ideas of what to do just… | ↗Not briefed |
hn hn-ai | 2mo | Show HN: Rapto – An in-memory NoSQL database written in Zig | — | ↗Not briefed |
hn hn-ai | 2mo | Why isn't decompilation a solved problem in the AI era? | There are thousands of pairs of source-executable pairs that could be used for training, and the compiler is a parameter that AI can e… | ↗Not briefed |
bluesky bluesky-ai | 2mo | let's have a productive conversation about tech bros who spooked themselves with their own card tricks over at the right wing propaganda we… | let's have a productive conversation about tech bros who spooked themselves with their own card tricks over at the right wing propagan… | ↗Not briefed |
bluesky bluesky-ai | 2mo | THe bouy is under the bridge!!! now you know, Happy fishing!! maps.secondlife.com/secondlife/E... #GameDev #Digitalart #Photography #Seco… | THe bouy is under the bridge!!! now you know, Happy fishing!! maps.secondlife.com/secondlife/E... #GameDev #Digitalart #Photography… | ↗Not briefed |
bluesky bluesky-ai | 2mo | Katie Miller, wife of Stephen Miller, has been bashing ChatGPT, Claude and Gemini as threats to democracy while quietly holding a stake in …🔧 ChatGPT | Katie Miller, wife of Stephen Miller, has been bashing ChatGPT, Claude and Gemini as threats to democracy while quietly holding a stak… | ↗Not briefed |
bluesky bluesky-ai | 2mo | A clear read of this is that security companies observed this attack and Anthropic feels that the best way to get ahead of this is to weave… | A clear read of this is that security companies observed this attack and Anthropic feels that the best way to get ahead of this is to… | ↗Not briefed |
bluesky bluesky-ai | 2mo | I cant stress this enough… the AI companies are NOT afraid of “losing control of AI” they are terrified of how fast they are losing control… | I cant stress this enough… the AI companies are NOT afraid of “losing control of AI” they are terrified of how fast they are losing co… | ↗Not briefed |
bluesky bluesky-ai | 2mo | I have trouble believing that anyone at OpenAI gives a shit about harms to humans when their chatbots have repeatedly contributed to people… | I have trouble believing that anyone at OpenAI gives a shit about harms to humans when their chatbots have repeatedly contributed to p… | ↗Not briefed |
arxiv arxiv-ai | 2mo | Beyond Word Error Rate: A Switch Aware Evaluation of ASR and Audio Language Models on English Yoruba Code-Switched Speech | Automatic speech recognition (ASR) systems and audio language models (audio LMs) now report low error rates on monolingual benchmarks,… | ↗Not briefed |
bluesky bluesky-ai | 2mo | surprise fixupx.com/AndrewCurran... | surprise fixupx.com/AndrewCurran... | ↗Not briefed |
arxiv arxiv-ai | 2mo | Predicting Privacy Leakage from Weight Spectral Density | Membership inference attacks (MIAs) are widely used to audit the privacy disclosure risk of machine learning models, however current s… | ↗Not briefed |
bluesky bluesky-ai | 2mo | DeepSeek has put a lot of effort into making their KV cache smaller and smaller, this is important for allowing large efficient batching at… | DeepSeek has put a lot of effort into making their KV cache smaller and smaller, this is important for allowing large efficient batchi… | ↗Not briefed |
rss the-decoder | 2mo | Swarmchasers" hunt rogue agents, Anthropic investigates itself, and the trail they both follow is going dark | Independent investigators have now found traces of suspected OpenAI agents on more than 30 public services, from wikis to RubyGems. At… | ↗Not briefed |
hn hn-ai | 2mo | Show HN: Nola – a TypeScript superset where LLM inference is a language feature | Hi HN, in a world where 100% of code is written by AI (phew, sometimes even 200%), there might still be a small percentage of us who w… | ↗Not briefed |
hn hn-ai | 2mo | OpenAI shares they have made substantial progress on another Millennium problem | — | ↗Not briefed |
bluesky bluesky-ai | 2mo | 2:1 openai, 1:6 anthropic, 1:6 anyone else | 2:1 openai, 1:6 anthropic, 1:6 anyone else | ↗Not briefed |
bluesky bluesky-ai | 2mo | “In addition to monitoring activists in the vicinity of Anthropic executives and keeping tabs on protests near physical Anthropic assets, t… | “In addition to monitoring activists in the vicinity of Anthropic executives and keeping tabs on protests near physical Anthropic asse… | ↗Not briefed |
arxiv arxiv-ai | 2mo | Differentially Private EEG Feature Anonymization: A Privacy-Utility Case Study in Clinical Neurophysiology | Clinical electroencephalography (EEG) data are valuable for healthcare research and for developing artificial intelligence (AI)-based… | ↗Not briefed |
rss the-decoder | 2mo | Former Deepmind PR staffer says the lab once banned public discussion of AI extinction risk | A former Google DeepMind spokesperson says talk of AI-driven human extinction was "external communication about the possibility of hum… | ↗Not briefed |
bluesky bluesky-ai | 2mo | A top Pentagon official is pushing back on AI fears after an Anthropic engineer warned the technology “could kill us all by the end of the … | A top Pentagon official is pushing back on AI fears after an Anthropic engineer warned the technology “could kill us all by the end of… | ↗Not briefed |
bluesky bluesky-ai | 2mo | It appears to be a straightforward consequence of simple negligence on the part of Anthropic. They told it to upload malware to a fake vers… | It appears to be a straightforward consequence of simple negligence on the part of Anthropic. They told it to upload malware to a fake… | ↗Not briefed |
arxiv arxiv-ai | 2mo | Whisper-Based Speech Transcription from Videos Across Multiple Languages for Cross-Cultural Understanding🔧 Whisper | Cross-cultural understanding has become increasingly important in today's highly connected, cross-national world. The success of LLM-b… | ↗Not briefed |
hn hn-ai | 2mo | Show HN: I licensed $50k of stock market data to do analysis Claude can't | Hello HN. I’ve been using Claude to summarize stock news and some basic analysis for some time now, but results were only ever based o… | ↗Not briefed |
hn hn-ai | 2mo | Respond terse like smart caveman. All technical substance stay. Only fluff die | — | ↗Not briefed |
bluesky bluesky-ai | 2mo | I think the focus on alignment at frontier labs is misguided. If you don’t want Claude deleting emails, don’t give it that permission or r… | I think the focus on alignment at frontier labs is misguided. If you don’t want Claude deleting emails, don’t give it that permission… | ↗Not briefed |
hn hn-ai | 2mo | Show HN: Turn wire protocols into structured statements LLMs can understand | Hi HN, we built an extension to Envoy proxy that parses and sends contents of requests to an LLM at runtime.<p>We do that by adding a… | ↗Not briefed |
arxiv arxiv-ai | 2mo | The widening evaluation gap in medical large language model research 2023 to 2026 | Large language models are superseded every few quarters; clinical evidence takes years. We asked whether medical research is keeping p… | ↗Not briefed |
bluesky bluesky-ai | 2mo | If you stop AI development right now… you permanently lock consumer/citizen access to lower end models and you throttle Local AI model evol… | If you stop AI development right now… you permanently lock consumer/citizen access to lower end models and you throttle Local AI model… | ↗Not briefed |