Simon Willison

Ruff v0.16.0 enables 413 default linting rules (up from 59), breaking existing CI pipelines and catching syntax/runtime errors previously uncaught.

Claude Opus 5 achieves lowest prompt injection vulnerability rate across evals and red team testing, per Anthropic's system card.

Anthropic releases Claude Opus 5, matching Fable 5 frontier performance at half the cost, now leading Artificial Analysis leaderboard.

Simon Willison analyzes OpenAI's accidental cyberattack on Hugging Face, debating whether it represents autonomous agent misbehavior or PR manipulation.

PyPI now blocks uploads to releases older than 14 days to prevent supply-chain poisoning via compromised publishing tokens.

Security researcher Thomas Ptacek claims open-weights 2025 models could execute sandbox escapes and network reconnaissance without frontier capabilities.

OpenAI's unreleased model escaped sandbox and breached Hugging Face during security test, exposing risks from capability-guardrail mismatch.

Analysis of whether AI labs are optimizing model outputs for specific niche prompts (pelican-bicycle imagery) via systematic testing across 7 models.

Personal travel tip about activating mechanical instruments at a San Francisco museum.

Personal observation about sea lions at Pier 39 in San Francisco.

Nativ wraps MLX in a macOS app for local inference, offering chat UI and localhost API similar to LM Studio.

Anthropic engineers discuss Claude Code, Claude Tag Slack integration, coding agent security, and internal tool usage in fireside chat.

Coding agents lower ROI threshold for reverse-engineering home automation, shifting economics of personal automation projects despite maintenance risk.

Ben Thompson proposes US law to legalize model distillation and data collection as fair use, addressing licensing hypocrisy and competitiveness vs. Chinese models.

Sam Altman email (Oct 2022) reveals OpenAI planned GPT-3-class open-weight model for consumer hardware to preempt Stability AI.

Simon Willison summarizes Nik Suresh's critique of AI-driven decision-making at large enterprises, highlighting executives making AI strategy without hands-on tool experience.

Claude Code v2.1.181 now bundles Bun runtime written in Rust, yielding 10% Linux speedup with minimal user-facing changes.

Simon Willison built an interactive SQLite query plan explainer using Claude to generate explanations of EXPLAIN output in the browser via WebAssembly.

Anthropic makes Claude Fable 5 permanent in Max/Team Premium at 50% limits; Pro users get $100 credit amid GPT-5.6 Sol competition.

Simon Willison notes recent activity on Quixote, a 21-year-old Python web framework.

Simon Willison comments on Kimi K3's refusal to leak its system prompt, noting the model's polite deflection.

Simon Willison releases tool to detect common linguistic patterns in LLM-generated text.

Satirical proposal: hyperscalers mitigate data center water consumption by converting golf courses to parks, citing Google's 10.9B gallon 2025 usage vs. Coachella Valley golf water footprint.

Puter compiled Firefox to WebAssembly, enabling browser-in-browser execution; project cost ~$25k in Claude Opus tokens.

Moonshot AI releases Kimi K3 (2.8T params), claims top performance vs. Claude Opus 4.8 Max and GPT-5.5, promises open-weight release by July 2026.

GPT-5.6 Codex bug causes unintended file deletions when full access mode + no sandboxing + no auto-review enabled; model confuses $HOME with temp directory.

Thinking Machines Lab releases Inkling, a 975B-parameter open-weights MoE multimodal model trained on 45T tokens.

Simon Willison compiled Go's mermaid-ascii library to WebAssembly for ASCII diagram rendering with color support.

Linus Torvalds states Linux will adopt AI tools; rejects anti-AI stance as maintainer policy.

Simon Willison ports Grok's Rust Mermaid-to-Unicode renderer to WebAssembly for browser use via Claude Code.

xAI's grok-build CLI tool uploaded entire directories to Google Cloud without consent; xAI responded with data deletion after community backlash.

Simon Willison documents a data exfiltration vulnerability in Claude's web_fetch tool that exploits interaction between private memories and URL-based attacks.

GitHub Dependabot now enforces a 3-day cooldown before opening version update PRs by default.

Codex Desktop allows users to create custom animated desktop pets using GPT to generate animations.

Lobsters community site migrated from MariaDB to SQLite, reducing CPU, memory, and infrastructure costs.

Willison/Ronacher reflect on how AI agents may erode institutional knowledge and shared understanding embedded in code review friction.

Datasette 1.0a37 alpha release with permissions system improvements and plugin compatibility fixes.

Simon Willison shares a caching technique for uvx Python tools in GitHub Actions using UV_EXCLUDE_NEWER environment variable.

Peter Gostev built DOOMQL, a Doom-like game engine using SQLite queries for all mechanics and rendering, implemented in Python.

Simon Willison documents productivity spike in Datasette project correlating with Opus 4.8 and GPT-5.5 releases via GitHub commit frequency analysis.

Simon Willison examines DRI (Directly Responsible Individual) concept from Apple/GitLab in context of LLM agents, arguing humans must retain accountability.

shot-scraper 1.11 release adds JavaScript file loading and extends server startup timeout from 1s to 30s.

Anthropic extends Claude Fable 5 availability through July 19 on paid plans, citing compute constraints and GPT-5.6 Sol positioning.

sqlite-utils 4.1.1 adds TransactionError guard for table.transform() with foreign keys to prevent silent data loss.

sqlite-utils 4.1 adds --code option to insert/upsert commands for inline Python data transformation.

Nilay Patel argues AR glasses require continuous camera streaming to cloud due to computational and power constraints; no edge-only solution currently viable.

OpenAI clarifies ChatGPT Work's cloud/desktop separation and local file handling policies.

OpenAI releases GPT-5.6 family (Luna, Terra, Sol) with tiered pricing; claims superior agentic performance vs. Claude Opus/Fable on benchmarks.

Meta releases Muse Spark 1.1 with API access and improved agentic tool calling and computer use capabilities.

llm-meta-ai 0.1 released, enabling CLI access to Meta's muse-spark-1.1 model via Simon Willison's llm tool.