The Archive

General-purpose VLMs remain unreliable for biomedical research because valid answers in scientific papers depend on evidence split across figures, tables, charts, captions, and referring text. Existing post-training pipelines are bottlenecked by costly expert annotation and by synthetic data that drops this evidence structure. We present Ryze, a fully automated system that converts raw biomedical papers into an evidence-enriched training set and a domain-specialized VLM. Ryze synthesizes QA pairs with complete supporting evidence (visual element, caption, extracted structure, and referring pa...

The goal, the firm said, is to develop and operate up to 5 gigawatts of additional data center capacity.

How we contain Claude across products A complaint I often have about sandboxing products is that they are rarely thoroughly documented , and in the absence of detailed documentation it's hard to know how much I can trust them. Anthropic just published a fantastic overview of how their various sandbox techniques work across Claude.ai , Claude Code, and Cowork. We constrain where and how an agent can act with process sandboxes, VMs, filesystem boundaries, and egress controls. The goal is to set a hard boundary on what an agent can reach. For example, if credentials never enter the sandbox, they...

Large language models systematically hallucinate legal citations -- fabricating statute references, citing repealed provisions, and confusing jurisdictions -- yet no automated method exists to measure or reduce this behavior at scale. We propose citation grounding (CG), a metric that verifies LLM-generated legal citations against a ground-truth citation graph extracted from 100.8 million Ukrainian court decisions (502 million edges, 21,736 unique statute nodes). CG decomposes into three components -- citation precision (does the cited provision exist?), citation relevance (is it contextually ...

This paper presents a fast, recursive neural solver for the J2-perturbed Lambert problem based on Tiny Recursive Models (TRM), termed the TRM-Perturbed Lambert (TRM-PL) model. TRM is a weight-shared architecture whose effective capacity emerges from iteration depth rather than parameter count: a compact reasoning module is applied repeatedly within a two-level latent hierarchy, refining a candidate departure velocity by simulating the J2 trajectory and correcting it from the resulting tracking error. This unifies initial-guess generation and iterative correction in a single, end-to-end differ...

Research: Running Python ASGI apps in the browser via Pyodide + a service worker Datasette Lite is my version of Datasette that runs entirely in the browser using Pyodide in WebAssembly. When I first built it four years ago I used Web Workers and code that intercepts navigation operations and fetches the generated HTML by running the Python app. This worked, but had the disadvantage that any JavaScript in <script> tags would not be executed - breaking some Datasette functionality and a whole lot of Datasette plugins. This morning I set Claude Opus 4.8 the task (in Claude Code for web) of figu...

The treatment of ischemic stroke using mechanical thrombectomy involves difficult decisions under intense time constraints. Numerical physics simulations can in theory inform operators to make better decisions regarding treatment approaches and device selection, but are too slow to do so in practice. In this thesis, we investigate if current machine learning based surrogates can accurately emulate these simulations in a step-by-step manner while making them significantly faster. To do this we train three surrogate models on two simulations that involve a simplified aspiration procedure, with ...

Phishing attacks remain a major cybersecurity threat, exploiting deceptive URLs to steal sensitive user information. Traditional blacklist and rule-based detection approaches are reactive and often fail to identify newly emerging phishing URLs. This paper proposes a lightweight hybrid framework for real-time phishing URL detection that combines blacklist-based screening with a Multi-Layer Perceptron (MLP) classifier operating solely on structural URL features. The framework extracts 16 URL-derived features capturing structural, domain-based, and security-related characteristics without requir...

I Am Retiring from Tech to Live Offline I've seen a lot of posts on forums from people threatening to quit their careers over AI. This is not one of those: Chad Whitacre is taking concrete steps, starting with this typewritten, scanned letter I'm retiring from tech. Well, "retiring" is euphemistic. I'm stepping away from tech, and that includes Open Source. [...] AI was the last straw. Have you heard of that island off India where the indigenous population kills any outsiders fool-hardy enough to land? They are doing the rest of us a favor by preserving a way of life we may need again someday...

My take on AI is, essentially, everybody who’s against it is too against it and everybody who’s for it is too for it. &mdash; Daniel Jalkut , via John Gruber Tags: ai

The golden age of Microsoft's Github Copilot appears to be at an end.

Meta seems to be making big bets on AI-powered hardware.

Gemini Spark helps automate everyday tasks, from inbox summaries to local event planning, but it’s unclear why Google made it a separate product.

"If you're 22 years old in San Francisco and building something in AI, there may be a seed term sheet in your inbox — but if you're 19, oh my God, this means you're really good; you might already have a Series A [offer]," said one, half-kiddingly.

We’ve compiled an overview of some of the top alternative browsers available today aiming to challenge Chrome and Safari.

A good time with old maps. | Image: Past Maps Craig Campbell walked away from the river of investor money flowing into AI to create, of all things, a website. Sure, Campbell probably could have started an AI company. He's a former engineer at Meta and an experienced tech founder who in 2022 sold his last venture - an e-commerce tool for businesses that use Shopify - right as the AI boom was booming. "I had my prior VC investors breathing down my neck, going 'start something else. We'll write you a blank check.'" He had other ideas. People generally aren't rushing to get into the website busin...

TikTok sellers that appear to be AI generated, in tears. | The Verge Aliyah, a light-skinned Black woman dressed in country-western gear, is struggling to sell metal buckles she handmade on TikTok. In a video for the social media platform from March, she cries to the camera and pleads for views: "Even as a black woman, I have more faith that white women will stay 13 seconds [on this video] to save my belt buckle business," the onscreen text reads. She wipes a tear off her cheek. But Aliyah isn't real, and neither are her supposedly handmade products - she's one of many AI-generated influencer...

Number go up? | Image: Cath Virginia / The Verge, Getty Images I haven't seen anything as stupid as the WeWork IPO document in a very long time - that is, until Elon Musk filed to take SpaceX public. WeWork was a joke. SpaceX is a threat. And if Musk and his bankers have their way, you are going to be their bagholder. Lots of the top-line details leaked long before the S-1 filing itself became public. There's the rumored valuation of more than $1 trillion. That's despite the nearly $5 billion in losses last year. The total addressable market (TAM) for SpaceX - the amount of revenue SpaceX thi...

a quiet day lets us highlight the new AIE WF focuses

Modern LLM serving is hard to tune because each deployment is a stack of interacting choices: model backend, tensor-parallel shape, prefill/decode split, worker... Modern LLM serving is hard to tune because each deployment is a stack of interacting choices: model backend, tensor-parallel shape, prefill/decode split, worker counts, scheduler settings, routing policy, KV cache behavior, autoscaling thresholds, and topology. Those choices interact across layers, and a local improvement can shift the bottleneck somewhere else. For larger models… Source

While AI is helping coders produce code faster, it may not be producing better code, researchers warn. And that could cause problems down the road for them.

We used Google AI Studio to vibe code a quiz about our top I/O 2026 announcements.

Connector-based video unified models have demonstrated strong capability in instruction-grounded video synthesis, but integrating a large high-fidelity generator into the unified training loop is computationally prohibitive, limiting achievable visual quality. We therefore propose Lumos-Nexus, a training-efficient unified video generation framework that facilitates the development of strong reasoning-driven generation capabilities while significantly enhancing visual fidelity. Lumos-Nexus adopts a two-stage design: 1) During training, only a lightweight generator is aligned with the understan...

Diffusion models have shown promising performance as data-driven priors for computational imaging, as well as some capacity to detect out-of-distribution (OOD) images. However, existing approaches to OOD detection often require some knowledge of the shifted distribution, fail to detect subtle or localized distribution shifts, and operate on full images, rather than the indirect measurements available in inverse problems. We propose an OOD detection metric based on the Kullback-Leibler divergence between the diffusion prior and the posterior distribution, that (i) does not require any calibrat...

Communication costs are a major bottleneck in distributed learning and first-order optimization. A common approach to alleviate this issue is to compress the gradient information exchanged between agents. However, such compression typically degrades the convergence guarantees of gradient-based methods. Error feedback mechanisms provide a simple and computationally cheap remedy for this issue, but numerous variants have been proposed, and their relative performance remains poorly understood. This paper provides tight convergence analyses for two of the main error-feedback algorithms from the l...

Language models can find thousands of severe software vulnerabilities, and agents are increasingly being misused for cyberattacks. To avoid detection, attackers frequently distribute their misuse, splitting a harmful task across many user accounts so each individual transcript looks benign. Because safety monitors score only one agent context at a time, they are structurally blind to misuse that is only visible in aggregate, across many accounts. We show this gap is real by building, to our knowledge, the first distributed agent attack, a multi-agent scaffold that completes hard cybersecurity...

Text-to-video (T2V) generation faces challenging questions when generating videos with long horizons containing multiple events. Inspired by the intrinsics of the diffusion process, we probe video diffusion transformers (DiTs) and uncover intrinsic turning points in the DiT denoising trajectory where conditioning text affects generation from global layout to fine-grained details. Building on this finding, we present TunerDiT, a simple yet effective progressive steering method that requires no additional training for multi-event generation. TunerDiT comprises two steering handles: (1) Event-Pa...

Grasping the semantics of rare constructions (form-meaning pairings) has been shown to be a challenging problem that has currently only been solved by the largest LLMs. It remains an open question if open-source models have robust constructional understanding, and if so, what learning dynamics underlie the acquisition of this knowledge. Focusing on a set of rare Paired-Focus constructions in English (e.g. "let alone", "much less"), we construct a novel dataset to test their meanings using both scalar adjectival semantics and general world knowledge. Testing a wide range of models differing in...

Long-context reasoning remains a central challenge for large language models, which often fail to locate and integrate key information in extensive distracting content. Reinforcement learning with verifiable rewards (RLVR) has shown promise for this task, yet existing methods are limited by low-confusability distractors and sparse, outcome-only reward signals that cannot supervise intermediate reasoning steps. To address these issues, we introduce \textsc{LongTraceRL}. For data construction, we generate multi-hop questions via knowledge graph random walks and leverage search agent trajectorie...

The same arguments often need to be evaluated under different external regimes. An agent with influence over the regime has a strategic lever that standard formalisms do not directly capture. We introduce context-dependent argumentation frameworks (CDAFs), an extension of Dung's theory in which a defeat function determines, per context, which attacks succeed. A perspective-labeled specialisation derives the defeat function from a relevance set $ρ$ and a priority $π$. The relevance set is the agent's action space. In a small worked example, the agent's target argument is rejected under every f...