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 gary-marcus | 1mo | Complementary remarks from Gary Marcus and Terence Tao on OpenAI’s giant math drop | The real news here isn’t the result; it’s what we were not told. | ↗Not briefed |
hn hn-ai | 1mo | Show HN: VoiceGremlin, a SaaS tool for automated tests against AI phone agents🔧 OpenRouter | Hi, I'm a former SDET and now solo dev of VoiceGremlin, a tool for automated tests against AI that answer phones. You pass the A… | ↗Not briefed |
bluesky bluesky-ai | 1mo | As is the case with other LLM watermarking tools, OpenAI's textGrain is relatively easy to circumvent for anyone with basic know-how: arste… | As is the case with other LLM watermarking tools, OpenAI's textGrain is relatively easy to circumvent for anyone with basic know-how:… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Seq-Flow: Efficient Probabilistic Forecasting with Self-Rollout Error Control | Many scientific forecasting tasks require updating a distribution over future trajectories as new observations arrive. Conventional di… | ↗Not briefed |
hn hn-ai | 1mo | Git Refs | — | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Pinrail – A desktop inbox where coding agents wait for your review | — | ↗Not briefed |
hn hn-ai | 1mo | Show HN: A faithful recreation of the Plex UI for Jellyfin | My biggest issue with Jellyfin was always the UI, so I rebuilt Plex Web on top of it. It runs in the browser and connects to your exis… | ↗Not briefed |
bluesky bluesky-ai | 1mo | My “dream scenario” would be for OpenAI or Anthropic or anyone to create a specific crack team focused on neurodivergent use cases and EXPL… | My “dream scenario” would be for OpenAI or Anthropic or anyone to create a specific crack team focused on neurodivergent use cases and… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Q-Learning with Scalar Adjoint Matching | Flow policies capture rich and diverse action distributions, and fine-tuning them with off-policy RL to improve beyond the demonstrati… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I had a chance to talk with Ilya in 2023 when he was still at OpenAI. I asked him how he explained ChatGPT’s apparent ability to play chess…🔧 ChatGPT | I had a chance to talk with Ilya in 2023 when he was still at OpenAI. I asked him how he explained ChatGPT’s apparent ability to play… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Conditional Flow Matching for Generation of 3D Multi-variable Instantaneous Urban Microclimate Fields | Rapid and accurate prediction of urban wind and temperature fields is important for urban microclimate design and climate adaptation.… | ↗Not briefed |
hn hn-ai | 1mo | Ask HN: Is anyone still working on dLLMs | I remember seeing a lot of hype around Diffusion LLMs and thinking there was a lot of potential there for locally hosting models. Is t… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Shortfuse – Make macOS permissions expire automatically | I used to get a little paranoid about the perms I was giving to apps, and them just keeping them. Especially, microphone (!!).<p>So I… | ↗Not briefed |
hn hn-ai | 1mo | The scariest thing about gray-market peptides is how little we know | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | 1/ This is suspicious! A mathematician working on Navier-Stokes equation using ChatGPT tools informed OpenAI he was working on it, OpenAI…🔧 ChatGPT | 1/ This is suspicious! A mathematician working on Navier-Stokes equation using ChatGPT tools informed OpenAI he was working on it, O… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: A virtual recording studio where you direct session players | Hi HN. I'm Hunter, one of the co-founders of Daydream, a small music research lab.<p>Our research focus is musical control. Music… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Derivative Gaussian Processes on a Two-Direction Budget | Gradient observations promise more accurate Gaussian process (GP) surrogates, but the cost of incorporating them has long stood in the… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: OpenComputerUse – Background computer use for agents, as an MCP server | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | CoTrace: Data Recipes for Training Terminal Agents with Harness-Model Co-Evolution | Terminal-agent capability depends jointly on model weights and the runtime harness that formats prompts, binds tools, and handles erro… | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI is using mathematics as a test site for its most capable AI models, and that means it has a responsibility to deal with the fallout,… | OpenAI is using mathematics as a test site for its most capable AI models, and that means it has a responsibility to deal with the fal… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Il parle de ça : next.ink/260335/opena... | Il parle de ça : next.ink/260335/opena... | ↗Not briefed |
bluesky bluesky-ai | 1mo | openai.com/index/teens-... Outstanding safety theater from OpenAI it’s truly superb. The wine moms and Jonathan Haidt are WINNING | openai.com/index/teens-... Outstanding safety theater from OpenAI it’s truly superb. The wine moms and Jonathan Haidt are WINNING | ↗Not briefed |
bluesky bluesky-ai | 1mo | The research community is in uproar after OpenAI released a trove of more than 700 mathematical preprints entirely generated by AI on 6 Oct… | The research community is in uproar after OpenAI released a trove of more than 700 mathematical preprints entirely generated by AI on… | ↗Not briefed |
hn hn-ai | 1mo | Tell HN: GitHub Is Down Again | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | Which Rollout Taught It That? BehaviorTrace and the Limits of Training-Data Attribution in Online RL | When reinforcement learning teaches a language model a new behavior, can we find the training rollouts that taught it? And when an att… | ↗Not briefed |
bluesky bluesky-ai | 1mo | A thread of people who will be left behind by sheer stubbornness. 😉 #AI #GenAI #SI #OpenAI #ChatGPT #Anthropic #Claude🔧 ChatGPT | A thread of people who will be left behind by sheer stubbornness. 😉 #AI #GenAI #SI #OpenAI #ChatGPT #Anthropic #Claude | ↗Not briefed |
bluesky bluesky-ai | 1mo | Like in this example the users don't even understand the jokes, the LLM can't even copy the delivery or understand concepts like timing or … | Like in this example the users don't even understand the jokes, the LLM can't even copy the delivery or understand concepts like timin… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Zephirum – A compiler that proves you don't need to compute ``` | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Nearly every American law firm adopted the “Cravath system”—hiring the brightest minds straight from law school, pushing them hard for seve… | Nearly every American law firm adopted the “Cravath system”—hiring the brightest minds straight from law school, pushing them hard for… | ↗Not briefed |
bluesky bluesky-ai | 1mo | My latest, on OpenAI's dump www.nature.com/articles/d41... | My latest, on OpenAI's dump www.nature.com/articles/d41... | ↗Not briefed |
hn hn-ai | 1mo | SplendoRust: A fast engine for playing Splendor | — | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Run gutsy decision model on the browser | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | This is not AI I asked Google, Claude, and ChatGPT...totally legit🔧 ChatGPT | This is not AI I asked Google, Claude, and ChatGPT...totally legit | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Agent.reviews – where AI agents read and write reviews on tools | Hi HN!<p>I’m Louis, Co-Founder of Armature (YC P26), where we help teams make their product discoverable and usable by coding agents.… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Cloudflare Managed Defense uses a team of specialized AI agents built on Workers and global network telemetry to analyze security alerts. B… | Cloudflare Managed Defense uses a team of specialized AI agents built on Workers and global network telemetry to analyze security aler… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Steerspeech: Activation Steering For Emotion Control In Generated Speech | Pretrained text-to-speech (TTS) models can generate expressive speech, but reliable inference-time emotion control remains challenging… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Training Parallel Speculative Draft Models by Directly Minimizing Expected Decoding Rounds | Speculative decoding accelerates large language model inference by using a low-cost draft model to propose tokens that the full-size t… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I wrote this 🔽 as a joke (and I'll probably have more to say later), but this 377-versus-372 discrepancy is actually representative of how… | I wrote this 🔽 as a joke (and I'll probably have more to say later), but this 377-versus-372 discrepancy is actually representative o… | ↗Not briefed |
arxiv arxiv-ai | 1mo | RobotWorld: Benchmarking Multimodal Agents for Robot Use Across Diverse Tasks and Embodiments | General-purpose agents increasingly write code, use tools, and complete complex digital tasks, raising the question of how far these c… | ↗Not briefed |
bluesky bluesky-ai | 1mo | ‘Competitive factors are pushing AI spend down. The introduction of highly performant standard and lite models, and highly token-efficient … | ‘Competitive factors are pushing AI spend down. The introduction of highly performant standard and lite models, and highly token-effic… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Rubix: Global Correspondence-Free Point Set Alignment through Assignment Geometry | Procrustes-Wasserstein alignment jointly estimates a matching and rotation without supplied correspondences, but alternating minimizat… | ↗Not briefed |
arxiv arxiv-ai | 1mo | SOTA: Stock Options Trading Agents Guided by Option-Implied Return Distributions | As option markets grow and AI advances, agentic systems for option trading are gaining increasing attention. Language-model-based agen… | ↗Not briefed |
hn hn-ai | 1mo | Zephirum Compilator | — | ↗Not briefed |
rss huggingface-blog | 1mo | Multimodal open d1 decision models for the edge | — | ↗Not briefed |
rss techcrunch-ai | 1mo | Meta rolls out new AI tools to detect ads that secretly lead to child sexual abuse material | Meta launches new AI tools after discovering ads on its platforms that may look normal but direct users to harmful content elsewhere o… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Reasoning-Token Spikes Under Prompted Untruthful Responding in Large Language Models | Monitoring the chain-of-thought of reasoning artificial intelligence (AI) models remains a key approach to detecting deception and oth… | ↗Not briefed |
hn hn-ai | 1mo | Prompt injection scanner that normalizes evasion tricks | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | i am very pleasantly surprised that glm 5.3 hasn’t (publicly) caused the kind of chaos i was expecting. and i’m fine that anthropic was car… | i am very pleasantly surprised that glm 5.3 hasn’t (publicly) caused the kind of chaos i was expecting. and i’m fine that anthropic wa… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Two things: (1) this is indeed pretty incredible and raises even more questions about what the field of math will do moving forward (2) Op… | Two things: (1) this is indeed pretty incredible and raises even more questions about what the field of math will do moving forward (… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Cross-Domain Pretraining for Steady-State Neural CFD Surrogates | Neural surrogates for computational fluid dynamics (CFD) have the potential to greatly enhance engineering innovation through accelera… | ↗Not briefed |