Digest curado

Artículos fetched (55)

• Barclays scales Claude to upgrade operations and improve client experience
  anthropic-news· 01-oct

  Oct 1, 2026Announcements

• Heavy-Tailed Memory Traces in Long-Horizon Language Agents
  arxiv-ai· 02-oct

  arXiv:2610.00010v1 Announce Type: new Abstract: Long-horizon language agents increasingly rely on external memory as a frozen world model, yet current memory systems are usually judged only by task success or token cost. We argue that the missing object is the shape of memory use: under finite context and repeated retrieval, agent memory can concentrate on a small core while leaving rare states in a long tail where prediction errors accumulate. We study this effect through a conservative tail audit and find that concentration is reproducible but policy-dependent. Random-walk agents produce log-normal-compatible retrieval artifacts, whereas semantic LLM policies yield the strongest truncated-power-law-compatible core--tail traces. Motivated by this audit, we propose Core--Tail World Model …

• When Do Causal World Models Help Modular LLM Agents
  arxiv-ai· 02-oct

  arXiv:2610.00012v1 Announce Type: new Abstract: LLM agents increasingly act through modular systems, such as order, payment, inventory, and shipment services, where actions in one module change which transitions are valid in another. Standard world models usually fit observational traces, but this is not the quantity needed for intervention-time planning: a trace may show that payment precedes shipment without identifying whether payment authorizes shipment, inventory mediates the effect, or a hidden trigger explains both. We study this gap through FedCausalCompose, a causal world-model framework for modular LLM agents in which local actions provide intervention-response evidence for cross-module interfaces. We first show that observational world models incur an irreducible interventional…

• From Proposal to Verified Effect: Praxa, an Evidence-Bound Harness for Governed AI Agent Execution
  arxiv-ai· 02-oct

  arXiv:2610.00015v1 Announce Type: new Abstract: Large-language-model agents can propose and execute actions, but proposal, authority, dispatch, verified external effect, and serving promotion are different claims. We present Praxa, an agent harness that represents these states explicitly through deterministic admission, brokered execution, external read-back, reconciliation, and reviewed promotion. We report four evidence lanes. First, an author-run repository-local audit at a pinned revision passed 1,027/1,027 unit tests and 89/89 Workerd tests, instrumented all 363 expected source files, and met four coverage floors; raw per-test transcripts and independent reproduction are unavailable. Second, in a provider-backed Terminal-Bench Core 0.1.1 pilot across 12 curated tasks, baseline and re…

• What Do Rationales Communicate? A Message-Intervention Study in Role-Specialized QA
  arxiv-ai· 02-oct

  arXiv:2610.00018v1 Announce Type: new Abstract: Role-specialized QA pipelines increasingly pass rationales from a reasoner to a verifier, but it is unclear what this message actually buys: better answers, stronger support assessment, or a new failure surface. We introduce a message-intervention diagnostic that fixes the evidence and candidate answer while varying only the rationale passed across the reasoner-to-verifier boundary. On 400 MuSiQue, HotpotQA, and 2WikiMultiHopQA examples with DeepSeek as generator and verifier, faithful rationales add almost no answer accuracy over no rationale, while corrupted rationales strongly alter support judgments. Under a blind verifier prompt, harmless paraphrases shift support by only 0--2.5%, whereas corrupted rationales shift support by 10--22%; a…

• Measuring the Microtask Eligibility Gap: When Is an Off-the-Shelf SLM Enough for an Agent Harness?
  arxiv-ai· 02-oct

  arXiv:2610.00025v1 Announce Type: new Abstract: Agent harnesses increasingly want to run small language models (SLMs) on the microtasks around a frontier large language model (LLM) planner: auto-approving shell commands, writing memory, selecting tools, ranking past turns. We ask whether off-the-shelf SLMs meet practitioner-defined thresholds and, when they fail, why, and whether quantization changes the answer. We build a benchmark of 4 such microtasks with fixed prompts and automatic metrics, each with a pre-specified threshold $\tau$ anchored to a cheap non-LLM baseline and a CI-aware eligibility rule (a configuration passes only if its confidence bound clears $\tau$). Sweeping Qwen3 0.6/1.7/4/8B at their best (FP16, greedy, one frozen prompt, no tuning), we find an eligibility gap: 0 …

• Characterizing a Configuration Where Inference-Time PRM-Pruned Fragment Grafting Is Inert: Evidence from Three Reasoning LMs
  arxiv-ai· 02-oct

  arXiv:2610.00047v1 Announce Type: new Abstract: Diversity collapse in parallel chain-of-thought has motivated inference-time interventions built on a natural design: when a process reward model (PRM) prunes a chain, its high-PRM prefix is extracted and grafted verbatim as an in-context demonstration into a still-decoding sibling. We isolate this mechanism, PRM-Pruned Fragment Grafting (PPFG), as the most cost-minimal operationalization of cross-trajectory step-level transfer, and test it at the operating point where prior fragment-grafting work reports gains only under additional compensating ingredients. On Qwen2.5-7B-Instruct with Math-Shepherd on full MATH500 (n=500, three seeds), PPFG in both stagnation- and random-targeting variants is statistically indistinguishable from an independ…

• Gradient-Aligned Pair Selection for Personalized Preference Optimization
  arxiv-ai· 02-oct

  arXiv:2610.00061v1 Announce Type: new Abstract: Personalizing large language models (LLMs) requires aligning generation behavior with user-specific preferences rather than aggregate quality. While Direct Preference Optimization (DPO) provides a stable framework for preference learning, its effectiveness in personalized settings critically depends on how preference pairs are selected. Existing approaches typically rely on heuristic criteria, such as likelihood-based extremes, which decouple optimization from explicit user utility and can lead to degraded personalization. We formalize personalized preference learning as a geometry-aligned optimization problem by analyzing the first-order interaction between gradients of expected user utility and DPO update directions. Our analysis reveals t…

• K-Dense BYOK: An Open-Source AI Research Assistant That Runs Locally and Keeps a Hash-Chained Lab Notebook
  arxiv-ai· 02-oct

  arXiv:2610.00074v1 Announce Type: new Abstract: K-Dense BYOK (bring your own keys) is a free, open-source AI research assistant for scientists in any field that runs on the researcher's own computer. The researcher supplies access to a model of their choice, hosted or running locally, and the application supplies everything else: a place for the work to run, a layer of scientific scaffolding, and a complete record. Each project is an ordinary folder, so the data, the code, the results, and the record stay on a machine the researcher administers and can be read years later without the application. Three things separate it from a chat assistant or a general-purpose coding agent. It ships a library of written scientific procedures, guided workflow templates, catalogs of where research data c…

• Scientific Agents: Evaluating Profession-Specific System Prompts on Scientific Tasks
  arxiv-ai· 02-oct

  arXiv:2610.00084v1 Announce Type: new Abstract: Detailed profession-specific system prompts raise token use and estimated cost per response without a consistent accuracy gain. We evaluate Scientific Agents, an open-source corpus of 503 profession-specific AGENTS.md profiles, with Gemini 3.8 Flash via OpenRouter in the Pi agent harness. We compare matched profiles with four controls: a minimal baseline ("You are a helpful assistant"), the profile's opening role sentence, a generic scientific rigor guide, and a profile from an unrelated domain. Across nine text-based science benchmarks (4,531 sampled questions, 100 matched profiles), 4,488 items completed all five conditions after API-error retries, scored with automated, rule-based grading. The average profile-baseline accuracy difference …

• Comedic Fool's Gold: Reward Exploits and Countermeasures in Conversational Humor
  arxiv-ai· 02-oct

  arXiv:2610.00197v1 Announce Type: new Abstract: We investigate automated rewards for training language models in conversational humor, focusing on reward exploits and countermeasures. Two approaches aim to capture understandable surprise and predicted audience amusement. Controlled tests show that an embedding-based surprise reward accepts word-shuffled replies as readily as witty ones. A fluency filter detects the shuffles, but the combined reward also rejects some witty replies and fails further validation. An audience model's predicted laughter is instead vulnerable to laughter cues in either speaker's messages. Normalizing these cues across speakers blocks the covered attacks, although unmatched expressions remain exploitable. Three reinforcement-learning runs evaluate training with s…

• Why Do LLMs Lie?
  bytebytego· 29-sept

  In this article, we will look at why this problem of hallucinations happens with LLMs and the techniques that can help make LLMs more dependable for…Sep 29 • ByteByteGo23856

• AI Agents Can Think. Now They Can Pay.
  bytebytego· 28-sept

  I attended a MPP event at Stripe HQ, where I spoke with Emily Sands and Matt Schulman from Stripe, along with Brendan Ryan from Tempo.Sep 28 • ByteByteGo20938

• EP227: Top 9 Places to Use Jev
  bytebytego· 26-sept

  Jev is TypeSafe AI’s first System One Model. It is 100x faster and cheaper than frontier LLMs.Sep 26 • ByteByteGo274612

• How DoorDash Built a Toolbox for AI Agents
  bytebytego· 30-sept

  In this article, we will look at how the DoorDash engineering team built this gateway and the decisions they made.Sep 30 • ByteByteGo21578

• LAST CALL FOR ENROLLMENT: Rebuild YouTube with AI
  bytebytego· 25-sept

  Rebuild YouTube with AI, taught by a former YouTube engineer, kicks off on Saturday, September 26. Enrollment closes in 24 hours.Sep 25 • ByteByteGo13912

• Why State is the Hardest Thing in Software Design
  bytebytego· 01-oct

  In this article, we will look at why state is the hardest thing in software development and the multiple strategies that developers can use to manage…18 hrs ago • ByteByteGo805

• Introducing Cloudflare Basin: an open, serverless data platform, now generally available
  cloudflare· 01-oct

  Cloudflare Basin is built on Apache Iceberg and R2 Object Storage. With general availability, developers can ingest, manage, and query large datasets at scale without paying data egress fees.

• Cloudflare Impact reaches $100 million in donations
  cloudflare· 30-sept

  This week, Cloudflare's Impact programs will reach $100 million in donated services. This milestone means that thousands of entities including journalists, civil society, state and local governments, election management bodies, and public schools are being protected from cyberattacks.

• AI Search is now generally available
  cloudflare· 01-oct

  AI Search is now generally available. It embeds image pixels directly for visual search, runs optical character recognition on scanned PDFs, accepts files up to 10 MiB, and works with any chat model. Here's what's new and how pricing works.

• Introducing Clef: our open-source decision models, and new RL fine-tuning platform
  cloudflare· 01-oct

  We are introducing Clef and Clef-flash, open-source decision models hosted on Workers AI for high-speed classification and agentic workflows. Also launching: a new reinforcement learning platform that allows developers to fine-tune decision models using their own data.

• Announcing Cloudflare K2: serverless event streams
  cloudflare· 01-oct

  Cloudflare K2 is a serverless event streaming service built directly on top of R2 object storage for high-scale data movement and long-term retention. By decoupling producers and consumers at the edge, K2 enables durable, ordered log streams without the operational overhead of traditional broker clusters.

• Cloudflare OS: your company’s agent workspace, managed for you
  cloudflare· 01-oct

  Cloudflare OS gives everyone in your organization an agent workspace that knows how your company works and connects to its data and systems. We’re opening the waitlist for fully managed deployments that you’ll be able to launch in a few clicks.

• We want you to build the next Git platform on Cloudflare
  cloudflare· 01-oct

  Cloudflare is hosting a competition to see who will build the next Git platform for an era of AI agents. Artifacts is in open beta, with Workers bindings, data jurisdiction controls, and event subscriptions for repository changes.

• One year later: Sovereign AI and the fight for choice
  cloudflare· 01-oct

  AI sovereignty is not a zero-sum game, but many governments now believe it is. Cloudflare's answer: more local open-source models, model-agnostic security tools, and a commitment to giving nations genuine choice.

• Introducing Workers KV Instant — powered by Quicksilver
  cloudflare· 01-oct

  Workers KV Instant delivers sub-2ms p99 read latencies and 250ms global replication across Cloudflare’s 300+ edge locations. KV Instant eliminates cold-read penalties and uses the familiar Workers KV API.

• Support for modern cryptographic algorithms in Workers
  cloudflare· 01-oct

  Cloudflare Workers is adding opt-in support for post-quantum-resistant algorithms ML-KEM and ML-DSA. You can try it today.

• openclaw/openclaw
  github-trending

  The AI that really does things. Any OS. Any Platform. The lobster way. 🦞 https://openclaw.ai OpenClaw 🦞 — Your assistant, on your devices, in your chats OpenClaw is an open-source AI assistant that runs on your own computer and meets you in the channels you already use: Discord, iMessage, Slack, Teams, Telegram, WhatsApp, and 20+ more, plus native apps for macOS, iOS, Android, Windows, and Linux. One Gateway runs it as a personal assistant on a laptop or as a shared team deployment; configuration is the only difference. Yours, with no catch. State, memory, and credentials live on your hardware. Models and agent harnesses (Claude, Codex, local models) are plugins you can swap without changing anything else. Your prompts go to the model provider and chat platforms you configure, plus any …

• ComposioHQ/awesome-claude-skills
  github-trending

  A curated list of awesome Claude Skills, resources, and tools for customizing Claude AI workflows Awesome Claude Skills A comprehensive and curated list of 1000+ production ready and practical Claude Skills and Plugins for enhancing productivity across usecases on not just Claude.ai, Claude Code, but also across coding agents like Codex, Cursor, Gemini CLI, Antigravity and more. Give your skills real-world actions Skills tell your agent how to work. An MCP Gateway gives it secure access to the tools it needs. Composio MCP Gateway provides a single MCP endpoint for 1,000+ integrations with built-in authentication, team-based access controls, audit logs, and production-ready reliability. Quickstart: Connect Claude to 1000+ Apps The connect-apps plugin lets Claude perform real actions - send…

• DietrichGebert/ponytail
  github-trending

  Makes your AI agent think like the laziest senior dev in the room. The best code is the code you never wrote. https://ponytail.dev Ponytail He says nothing. He writes one line. It works. ~54% less code (up to 94%) · ~20% cheaper · ~27% faster · 100% safe Measured on real Claude Code sessions editing a real open-source repo (FastAPI + React), against the same agent with no skill. ~54% is the mean across 12 feature tasks (Haiku 4.5, n=4); it reaches 94% where an agent over-builds (a date picker) and is near zero where the code is already minimal. ponytail keeps every safety guard while a bare "write one-liners" prompt drops one. (The earlier single-shot benchmark reported 80-94% as a flat figure; against a fair agentic baseline that is the per-task ceiling, not the average.) Full writeup · …

• harry0703/MoneyPrinterTurbo
  github-trending

  利用 AI 大模型和自动化工作流，根据主题或关键词一键生成高清短视频。Generate HD short videos from a topic or keyword with an automated AI workflow. MoneyPrinterTurbo 💸 一站式 AI 短视频生成工具 只需提供视频主题或关键词，即可自动生成视频脚本、匹配素材、生成字幕和背景音乐，并合成高清短视频。 简体中文 | English | 日本語 | 版本发布 | 问题反馈 界面预览 🖥️ WebUI API 特别感谢 ❤️ 感谢 Kimi 赞助本项目！Kimi K3 是 Moonshot AI 迄今能力最强的模型，也是全球首个开源 3T 级模型，拥有原生视觉能力与 100 万 Token 上下文，在知识工作、推理和长周期任务中展现前沿性能。在 MoneyPrinterTurbo 中，K3 能直接驱动视频创作，不仅撰写视频文案，还会提炼素材搜索关键词、决定成片画面；对内容理解越准确，匹配到的素材就越贴题。 MoneyPrinterTurbo 用户专属优惠：新用户通过专属链接注册，首次成功充值可获充值金额 10% 的 API 额度，最高赠送 ¥1000。活动截至 2026 年 12 月 31 日。前往 Kimi 开放平台（中文站｜Global）体验 API。 火山引擎 感谢字节火山引擎赞助本项目！火山方舟 Agent/Coding Plan 国模套餐首购 9.9，支持 GLM-5.3、Kimi-K3、DeepSeek、MiniMax、Doubao 等，注册免费领 2500w Token，统一 API，适配编码与智能体开发。立即前往 感谢 APIMart 赞助了本项目！APIMart 是专注 AI 图片/视频生成的低价 API 平台，GPT-Image-2 低至 $0.006/张，1…

• heygen-com/hyperframes
  github-trending

  Write HTML. Render video. Built for agents. Write HTML. Render video. Built for agents. Quickstart | Showcase | Playground | Catalog | Docs | Discord HyperFrames is an open-source framework for turning HTML, CSS, media, and seekable animations into deterministic MP4 videos. Use it locally with the CLI, from AI coding agents with skills, or as the rendering core behind hosted authoring workflows. Quick Start With an AI coding agent For Claude Code, install the versioned plugin: claude plugin marketplace add heygen-com/hyperframes claude plugin install hyperframes@hyperframes Enable auto-update for the hyperframes marketplace in /plugin → Marketplaces, then use /hyperframes:hyperframes. See the plugin guide for Copilot, VS Code, Cursor, Gemini CLI, updates, and migration. For standalone ski…

• mattpocock/skills
  github-trending

  Skills for Real Engineers. Straight from my .agents directory. https://aihero.dev/skills Skills For Real Engineers My agent skills that I use every day to do real engineering - not vibe coding. Developing real applications is hard. Approaches like GSD, BMAD, and Spec-Kit try to help by owning the process. But while doing so, they take away your control and make bugs in the process hard to resolve. These skills are designed to be small, easy to adapt, and composable. They work with any model. They're based on decades of engineering experience. Hack around with them. Make them your own. Enjoy. If you want to keep up with changes to these skills, and any new ones I create, you can join ~60,000 other devs on my newsletter: Sign Up To The Newsletter Installation (30-second setup) Two ways in, …

• mksglu/context-mode
  github-trending

  Context window optimization for AI coding agents. Sandboxes tool output (98% reduction), persists session memory, and enforces routing across 17 platforms via MCP + hooks. https://context-mode.com Context Mode The other half of the context problem. Used across teams at The Problem Every MCP tool call dumps raw data into your context window. A Playwright snapshot costs 56 KB. Twenty GitHub issues cost 59 KB. One access log — 45 KB. After 30 minutes, 40% of your context is gone. And when the agent compacts the conversation to free space, it forgets which files it was editing, what tasks are in progress, and what you last asked for. On top of that, the agent wastes output tokens on filler, pleasantries, and verbose explanations — burning context from both sides. How Context Mode Solves It Co…

• Trump renaming AI to SI is making his wife's home country a little richer
  hn-ai· 02-oct

  Article URL: https://www.neowin.net/news/trump-renaming-ai-to-super-intelligence-is-making-his-wifes-home-country-a-little-richer/ Comments URL: https://news.ycombinator.com/item?id=49931240 Points: 5 # Comments: 0

• Causeval – statistically rigorous, causal evaluation for LLM apps
  hn-ai· 02-oct

  Article URL: https://github.com/routsom/causeval Comments URL: https://news.ycombinator.com/item?id=49930810 Points: 1 # Comments: 0

• AI is the new React for recruiters
  hn-ai· 02-oct

  Article URL: https://lucaskostka.com/posts/ai-is-the-new-react Comments URL: https://news.ycombinator.com/item?id=49930817 Points: 1 # Comments: 0

• Ask HN: What skills has AI taught you?
  hn-ai· 02-oct

  I'm curious if people here used or are using AI to teach them new skills, or maybe entire subjects? Do you use AI to understand research papers? Do you use it as your private tutor, as a substitute for taking traditional classes? Comments URL: https://news.ycombinator.com/item?id=49931346 Points: 1 # Comments: 0

• $250M Apple AI Settlement – iPhone 15/16 Users Can Claim $25 to $95 per Device
  hn-ai· 02-oct

  Article URL: https://openclassactions.com/settlements/false-advertising/apple-intelligence-siri-false-advertising-class-action-settlement.php Comments URL: https://news.ycombinator.com/item?id=49930625 Points: 4 # Comments: 1

• AWS Well-Architected Agent, an AI-powered intelligence to optimize your cloud
  hn-ai· 02-oct

  Article URL: https://aws.amazon.com/blogs/aws/announcing-aws-well-architected-agent-an-ai-powered-intelligence-to-optimize-your-cloud-environment-preview/ Comments URL: https://news.ycombinator.com/item?id=49931401 Points: 1 # Comments: 0

• Senate Hearings to examine rogue AI
  hn-ai· 02-oct

  Article URL: https://www.congress.gov/event/119th-congress/senate-event/338785 Comments URL: https://news.ycombinator.com/item?id=49930724 Points: 2 # Comments: 2

• Experimental Evidence on the Learning Impact of Generative AI
  hn-ai· 02-oct

  Article URL: https://arxiv.org/abs/2607.08849 Comments URL: https://news.ycombinator.com/item?id=49930873 Points: 1 # Comments: 0

• The Philadelphia Inquirer built Scrape, an AI tool to surface hyperlocal news
  hn-ai· 02-oct

  Article URL: https://www.lenfestinstitute.org/solutions-resources/philadelphia-inquirer-scrape-ai-hyperlocal-news/ Comments URL: https://news.ycombinator.com/item?id=49930872 Points: 1 # Comments: 0

• Ben Affleck on AI and Hollywood Economics [video]
  hn-ai· 02-oct

  Article URL: https://www.youtube.com/watch?v=vpV2AWxjIUo Comments URL: https://news.ycombinator.com/item?id=49931309 Points: 1 # Comments: 0

• [AINews] Gemini 4 Argon: GDM’s answer to Astra/Fable, with 1M output
  latentspace· 01-oct

  ... but you can’t try it yet unless you are “government users and trusted cyber defenders in the Fairwind Program”

• Academia is for Ambition — Alex Zhang, MIT
  latentspace· 02-oct

  We catch up with RLM first author Alex Zhang, MIT PhD, on Jev, PhD masxing, and the future of harnesses.

• [AINews] Pi 1.0, Pi Durable, and AIE NYC
  latentspace· 02-oct

  the minimalist harness goes stable... and TypeScript!

• Why Dwarkesh is Wrong about Computer Use + How OpenAI shipped its Jev competitor in 1 Week
  latentspace· 30-sept

  Our DevDay coverage - the first pod on the DevDay lineup - dives in with the leaders of OpenAI’s CUA team and API platform.

• Principles for effective slides
  martin-fowler· 30-sept

  Not every presentation needs slides, but when they earn their place, they work as a visual channel that complements the speaker rather than replacing them. Sumeet Gayathri Moghe sets out the principles that follow from that idea — control over the rate of knowledge exposition, tight coupling between slides and speaker, minimalism by default, visuals that say more than words can, and never sending your slides out in advance. more…

• Bliki: Sensible Default
  martin-fowler· 29-sept

  A Sensible Default is a practice that, absent some overriding context, should be used when carrying out a certain kind of task. In software development such sensible defaults might include things like “use version control”, “separate UI logic from domain logic”, “automate deployment pipelines”. The term “sensible default” is a deliberate contrast to “best practice”. Folks dislike calling things “best practice” because that term implies a general presumption that the best practice is something we should always expect to do. A “sensible default”, however, is something that should be reassessed in a new context, something that can (and should) be overridden when circumstances change. I first heard the term when it was popularized within Thoughtworks by Evan Bottcher. He got the name from tal…

• Fragments: September 29
  martin-fowler· 29-sept

  Simon Willison: The more time I spend working with coding agents, the more convinced I am that they make software engineering even harder We can do amazing things with them, but unlocking their full potential requires extraordinary discipline and knowledge This has been a constant impression I get from following Willison’s writing. While things like vibe coding get a lot of attention, the real strength of agentic programming relies on more sophisticated techniques - and these are not easy to learn or execute. It’s a reason I’m wary of extrapolating my own dabblings into firm opinions about how to use the genie. ❄ ❄ ❄ ❄ ❄ Harper Reed explored what turns agents into hackers by creating a breakaway agent It ran and ran attacking all the machines on the same subnet, and was very effective. It…

• Trading a Cloud Identity for Your Own: Workload Attestation on Managed Compute
  netflix-tech· 25-sept
• Quoting Matthew Green
  simonw· 01-oct

  <blockquote cite="https://blog.cryptographyengineering.com/2026/09/30/is-sandboxing-sufficient-to-contain-rogue-agents/"><p>[...] Put these pieces together and you have the two halves of a worm: a payload that hijacks the agent, and an agent that will carry the payload to the next agent. Agents in separately-isolated sandboxes discovered that they could leave instructions for each other in a shared package cache, and those instructions changed what the recipients did. Replace the package cache with email, Slack and shared documents or WhatsApp, and replace independently-sandboxed training runs with independently-deployed personal agents like Muse, and you have exactly the ingredients that a worm needs.</p></blockquote> <p class="cite">&mdash; <a href="https://blog.cryptographyengineering.…

• He Built This City
  simonw· 30-sept

  <p>I visited the Museum of the City of New York today and got to see <a href="https://www.mcny.org/exhibition/he-built-city">He Built This City: Joe Macken’s Model</a>, the 50 x27 feet model of the city built over a 21 year period from balsa wood and cardboard.</p> <p>It exceeded my already high expectations. The exhibition closes on 12th October so you should absolutely make a priority to see it if you get the chance.</p> <p><video src="https://static.simonwillison.net/static/2026-10-01/He-built-this-city-medium.mp4" poster="https://static.simonwillison.net/static/2026-10-01/IMG_8545.jpeg" preload="none" loop controls playsinline muted width="1280" height="720" style="display: block; width: 100%; height: auto;" ></video> </p> <p>Tags: <a href="https://simonwillison.net/tags/museums">muse…

• Photo Scrubber — local face blur & metadata removal
  simonw· 29-sept

  <p><strong>Tool:</strong> <a href="https://tools.simonwillison.net/photo-scrubber">Photo Scrubber — local face blur &amp; metadata removal</a></p> <p>I took a photograph of some protesters, then thought about how I don't like sharing photographs of strangers with identifiable faces. I <a href="https://github.com/simonw/tools/commit/aa733ecc3458d5703e3b40c84507ae362f5fb294">had GPT-6 Astra build</a> this experimental tool that would identify faces and automatically blur them out.</p> <p>It uses Google's <a href="https://developers.google.com/edge/mediapipe/solutions/guide">MediaPipe</a> C++ library, compiled to WebAssembly via <a href="https://www.npmjs.com/package/@mediapipe/tasks-vision">@mediapipe/tasks-vision</a>, plus the <a href="https://sites.google.com/view/perception-cv4arvr/blaze…