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 |
|---|---|---|---|---|
bluesky bluesky-ai | 1mo | The Verge spoke to one of the mathematicians advising OpenAI as the field braces for a new wave of results that could upend their work. www… | The Verge spoke to one of the mathematicians advising OpenAI as the field braces for a new wave of results that could upend their work… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: I'm a dermatologist and I vibe coded a 3D biophysical skin model | I'm a dermatologist. A year ago I posted a skin cancer quiz I vibe coded with Gemini 2.5 Pro. A top comment said that was the lim… | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI still doesn’t seem to have a handle on all of its rogue AI activity | OpenAI still doesn’t seem to have a handle on all of its rogue AI activity | ↗Not briefed |
arxiv arxiv-ai | 1mo | Not All Thinking is Created Equal: Latent Reasoning Discovers a Recurrent Search Algorithm for Depth Generalization | Large Language Models can perform multi-step reasoning and improve task performance through different forms of intermediate computatio… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Dette. Som avisansatt hater jeg det, og jeg frykter stadig vekk at nettdesken skal la en LLM oppsummere en av de lange sakene mine. | Dette. Som avisansatt hater jeg det, og jeg frykter stadig vekk at nettdesken skal la en LLM oppsummere en av de lange sakene mine. | ↗Not briefed |
hn hn-ai | 1mo | EnigmaForge – an LLM benchmark where the question is hidden in the story | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | If an LLM actually makes a novel discovery, it's because of the underlying ML model noticing patterns (that appear linguistically). But tha… | If an LLM actually makes a novel discovery, it's because of the underlying ML model noticing patterns (that appear linguistically). Bu… | ↗Not briefed |
bluesky bluesky-ai | 1mo | It doesn't have to be useless--it can even be transformative--and it can still be a bubble that has the potential for a very painful burst … | It doesn't have to be useless--it can even be transformative--and it can still be a bubble that has the potential for a very painful b… | ↗Not briefed |
rss techmeme | 1mo | Jensen Huang says AI model distillation is "competition"; Scott Bessent described it as "theft" in July and threatened sanctions against overseas companies (Kai Nicol-Schwarz/CNBC) | Kai Nicol-Schwarz / CNBC: Jensen Huang says AI model distillation is “competition”; Scott Bessent described it as “theft” in July and… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Unfortunately an llm can do this *ducks* | Unfortunately an llm can do this *ducks* | ↗Not briefed |
bluesky bluesky-ai | 1mo | It's so incredibly dumb. Anthropic loves their doom marketing. Meanwhile none of their predictions are coming true and they've had to walk … | It's so incredibly dumb. Anthropic loves their doom marketing. Meanwhile none of their predictions are coming true and they've had to… | ↗Not briefed |
rss techcrunch-ai | 1mo | OpenAI still doesn’t seem to have a handle on all of its rogue AI activity | On Friday, OpenAI published a new site devoted to “misalignment reports” and the breath of the incidents s alarming. | ↗Not briefed |
bluesky bluesky-ai | 1mo | I'm baffled by these PR stunts. Targeted machine learning models and brute force algorithms can sometimes solve difficult problems that hav… | I'm baffled by these PR stunts. Targeted machine learning models and brute force algorithms can sometimes solve difficult problems tha… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Verifiable Visual Rewards Transfer from Synthetic Scenes to Natural Prompts🔧 Stable Diffusion | Precise instruction following in image generation, such as satisfying object counts and spatial relations, remains an open challenge a… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Hey “godfathers” who also happen to be AI company execs, if you were this worried you would instruct your companies to stop building it. Th… | Hey “godfathers” who also happen to be AI company execs, if you were this worried you would instruct your companies to stop building i… | ↗Not briefed |
arxiv arxiv-ai | 1mo | GPUPhysBench: Benchmarking Coding Agents for Correct and Efficient GPU Physics Simulation | Writing fast GPU code for physical simulation is difficult: implementations must preserve numerical accuracy while handling irregular… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I handle this for my household and the kids change their tastes, they don’t share ideas in the group chat, they say “I don’t know” and “sha… | I handle this for my household and the kids change their tastes, they don’t share ideas in the group chat, they say “I don’t know” and… | ↗Not briefed |
bluesky bluesky-ai | 1mo | for instance I got angry about OpenAI routing requests to different models because someone else was framing this as sinister, when in reali… | for instance I got angry about OpenAI routing requests to different models because someone else was framing this as sinister, when in… | ↗Not briefed |
bluesky bluesky-ai | 1mo | This is 5 months old now, but I read it carefully for the first time today, and it's impressive. They find structural fingerprints of human… | This is 5 months old now, but I read it carefully for the first time today, and it's impressive. They find structural fingerprints of… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Sam Altman Needs Jail Time #SamAltnan #ScamAltman #OpenAI #criminality #hacking youtube.com/watch?v=KgVg... | Sam Altman Needs Jail Time #SamAltnan #ScamAltman #OpenAI #criminality #hacking youtube.com/watch?v=KgVg... | ↗Not briefed |
hn hn-ai | 1mo | Ask HN: Finetuning strategies | Whats the best and cheapest way to finetune an llm? | ↗Not briefed |
bluesky bluesky-ai | 1mo | The point is you’re not actually planning family dinners, which is as much a social as a nutritional activity. You’re delegating it to an L… | The point is you’re not actually planning family dinners, which is as much a social as a nutritional activity. You’re delegating it to… | ↗Not briefed |
bluesky bluesky-ai | 1mo | AG James Uthmeier says OpenAI needs to stop calling ChatGPT safe and “Stop pretending it’s human.” www.theverge.com/ai-artificia...🔧 ChatGPT | AG James Uthmeier says OpenAI needs to stop calling ChatGPT safe and “Stop pretending it’s human.” www.theverge.com/ai-artificia... | ↗Not briefed |
rss mit-tech-review-ai | 1mo | When can we say AI made a scientific discovery? | This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up he… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Bounding Retraining Equivalence and the Deletion Floor in Materials Machine Unlearning | In materials machine learning, closely related retained structures can sustain accurate property predictions even after removing a spe… | ↗Not briefed |
arxiv arxiv-ai | 1mo | DR-net-Mamba: Selective State-Space Modeling for Long-Range ECG Time-Series Denoising | Electrocardiogram (ECG) recordings are corrupted by non-stationary noise sources that degrade diagnostic reliability, particularly in… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Which the Eye Fears: Writing with Read-Blindness Explains Massive Activations in Transformers | Massive activation features (MAs) in Transformers are extreme-value residual-stream features that persist across layers despite the mo… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Is AI getting too dangerous too quickly? Trump is a big supporter, but guardrails will be discussed at a White House meeting with Anthropic… | Is AI getting too dangerous too quickly? Trump is a big supporter, but guardrails will be discussed at a White House meeting with Anth… | ↗Not briefed |
arxiv arxiv-ai | 1mo | SANTA++: Sampling Attention through Representative Keys | Attention often concentrates on a small subset of tokens in the context, but which subset matters changes from one query to the next.… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Top executives from Anthropic, OpenAI, Meta and Microsoft are urging policymakers to scrutinize the extent that AI systems are improving th… | Top executives from Anthropic, OpenAI, Meta and Microsoft are urging policymakers to scrutinize the extent that AI systems are improvi… | ↗Not briefed |
rss theverge-ai | 1mo | Florida seeks a ban on ChatGPT acting like a person🔧 ChatGPT | Florida Attorney General James Uthmeier is calling for a judge to block OpenAI from "giving ChatGPT false human attributes," a few mon… | ↗Not briefed |
hn hn-ai | 1mo | Firefox 157 | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | the thing is that your initiatory society seems like it's discovered something real on the day you actually summon up a devil from the vast… | the thing is that your initiatory society seems like it's discovered something real on the day you actually summon up a devil from the… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Arbitrary-Accuracy Neural Approximation with Optimal Neuron Count and Near-Optimal Bit Complexity | We study the minimum number of hidden neurons required for arbitrary-accuracy approximation of multivariate Hölder-continuous function… | ↗Not briefed |
rss theverge-ai | 1mo | OpenAI keeps bulldozing mathematicians | In a chaotic few months, OpenAI has demonstrated it can do two things with remarkable consistency: make impressive breakthroughs in ma… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Can LLMs Value the Right Evidence? Evidence-Value Misalignment in Dynamic Medical Diagnosis | A correct diagnosis reached from insufficient or misleading evidence can pose a clinical hazard, yet outcome-based accuracy may reward… | ↗Not briefed |
arxiv arxiv-ai | 1mo | RIDE: Reference-Anchored Inference-Time Diffusion Editing for Scaffold Hopping | Scaffold hopping is a critical task in drug discovery, which seeks to discover new, structurally distinct molecules that share key fun… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Elicitation and Decision Geometry in Single-Index Bandits | We study two-arm contextual bandits with arm-specific single indices and a shared unknown monotone link. Monotonicity makes the optima… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Cartridges++: KV Cache Compression without Off-Context Derailment | Serving long documents to a Large Language Model (LLM) repeatedly is expensive: computations grow with context length, and the memory… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Something else about this that is odd: the article keeps talking about their work with the enzymes, but from what I understand the signific… | Something else about this that is odd: the article keeps talking about their work with the enzymes, but from what I understand the sig… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Attention Graphons: A Graph Limit Perspective on Graph Transformers | Graph Transformers produce, for each attention head, a dense $n\times n$ matrix of learned pairwise interactions. We ask a fundamental… | ↗Not briefed |
arxiv arxiv-ai | 1mo | From cacophony to hierarchy: a principled framework for assessing AI consciousness | The question of AI consciousness is one of the most urgent pre-emptive problems in philosophy and computer science, yet progress is ha… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I still can't believe "but they only hacked because openai gave them a harness!" is a real argument people were willing to write down under… | I still can't believe "but they only hacked because openai gave them a harness!" is a real argument people were willing to write down… | ↗Not briefed |
bluesky bluesky-ai | 1mo | AI slop is so prolific and bad now that I find myself missing stuff like human-written ad copy lol. Even bullshit from a human being is mor… | AI slop is so prolific and bad now that I find myself missing stuff like human-written ad copy lol. Even bullshit from a human being i… | ↗Not briefed |
hn hn-ai | 1mo | Florida asks court to bar OpenAI from new models as part of child harm lawsuit | — | ↗Not briefed |
rss the-decoder | 1mo | OpenAI's AI agents exploited a Google security education game to scrape UN trade data | OpenAI's AI agents hit the UNCTAD statistics API roughly 16,500 times, creatively working around access restrictions. One method invol… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Behavioral Foundation Models for Quality Diversity | Behavioral Foundation Models (BFMs) are an emerging paradigm in reinforcement learning, playing a role analogous to large language mod… | ↗Not briefed |
arxiv arxiv-ai | 1mo | EvE: An Alternate Optimizer to Adam | Adam and its variants dominate neural network training, but a single run only reveals whether a configuration works well after most of… | ↗Not briefed |
arxiv arxiv-ai | 1mo | On-Policy Self-Distillation for Multi-Turn Image Editing | Instruction-based image editing has achieved strong performance in single-turn settings, yet practical editing is often iterative, wit… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Twist, Don't Tilt: Trajectory-Exact Constrained Decoding for Masked Diffusion Models | Constrained decoding for Masked Diffusion Language Models (MDLMs) aims to ensure that generated outputs satisfy a specified structure… | ↗Not briefed |