Simon Willison — Independent developer and writer documenting practical experiments, tools, and deep analysis around large language models, generative AI, web development, security, and emerging programming workflows through detailed posts and daily TILs.
381 articles from this blog
Announcement of micropython-wasm 0.1a1, a Python library for running MicroPython sandboxes using WebAssembly.
A tool for creating a text editor with automatic file attachment via paste, drag-and-drop, or file selection.
Announcing micropython-wasm 0.1a0, a Python library for running MicroPython in a WebAssembly sandbox via wasmtime.
Hackers exploited Meta's AI support bot to take over high-profile Instagram accounts by simply asking it to change linked emails.
Datasette 1.0a32 bugfix release: fixes INSERT...RETURNING queries and base_url issues for the open-source data exploration tool.
Explores the paradox of AI tools boosting productivity yet causing distraction, with personal anecdotes and ADHD perspectives.
A quote from Karen Kwok explaining how Anthropic defines 'run-rate revenue' using consumption and subscription models.
Overview of Anthropic's sandbox techniques for Claude across products like Claude.ai, Claude Code, and Cowork.
A quote from Daniel Jalkut on AI, shared by Simon Willison, expressing a balanced view on the technology.
Running Python ASGI apps in the browser using Pyodide and a service worker, eliminating backend servers.
Datasette 1.0a31 alpha release adds SQL write queries and stored queries for data exploration.
Anthropic's run-rate revenue hits $47 billion, as disclosed in their Series H announcement, showing rapid growth.
Anthropic releases Claude Opus 4.8 with modest improvements, emphasizing honesty and reduced hallucinations in AI models.
A tool to render markdown with live preview, supporting tables, code blocks, and interactive SVG from URLs or pasted content.
SQLite adds AGENTS.md file clarifying no acceptance of AI-generated code, while splitting AI bug reports into a separate forum.
Analysis of Anthropic and OpenAI achieving product-market fit through enterprise API pricing and rising AI usage costs.
A humorous anecdote about an AI-generated image of a horse riding an astronaut on a pelican on a bicycle, with a self-added sign.
A quote from Romain Huet confirming OpenAI won't release a separate GPT-5.5-Codex model, as coding capabilities are unified.
A guide on prompting GPT-5.5, including tips for multi-step tasks, migration advice, and starting fresh with new prompts.
Release notes for llm 0.31, a command-line tool for accessing large language models, featuring GPT-5.5 support and new options.
