Simon Willison
25 articles tagged with this topic
AI Wrote a Million Lines. The Hard Part Was the Scaffolding — Paul Dix
Via Simon Willison, Paul Dix: 1M lines of AI-written code shipped to millions of dev machines. The real challenge wasn't the AI — it was the scaffoldi
Simon Willison: Don't Review AI Code Line-by-Line — Direct and Verify
Simon Willison this week: AI code needs confident instruction and verification, not line-by-line review. An underrated Agent-era soft skill.
AI Coding Output 100x Faster — The New Bottleneck Is Cognition and Judgment
Simon Willison argues coding agents can multiply output, but the true scarcity becomes the cognitive capacity to understand, maintain, and trade off c
LLMs Turn Software Into Lego — Morrell: The Era of User-Written Plugins Is Here
Morrell: LLMs cut extension costs to near zero; sandbox tech solves safety. Future software = 'reliable core' + user-built AI extensions. Big implicat
Alibaba's Qwen 27B Earns Rare Praise — Local AI Models Are Now Genuinely Usable
Simon Willison praised Alibaba's Qwen 27B for running smoothly on personal laptops — another signal open-source is catching closed APIs.
No AI News This Week — We Read About a Bird Living 14 Years at a California Port
We scanned 40+ sources—no AI news worth writing. But Morris the gannet on Simon Willison's blog made us pause: what deserves recording is itself a jud
1,856 tags won't fit in a prompt? Let AI hallucinate, then vector-match
Tag library too big for a prompt? Let the LLM hallucinate, then vector-match. LLMs generate, vector search constrains.
An Australian Gym Booking System Just Got Broken by AI — And the API Security Problem Nobody's
Developer Simon Willison revealed: an Australian gym booking site's cancellation API has zero auth checks, letting AI tools bump you off a waitlist. N
Anthropic Bakes Export Controls Into Its System Prompt — AI Firms Turn Politics Into a Product
Claude Opus 5's system prompt includes a passage detailing its June 2026 US Commerce Department export restriction and July reinstatement — signaling
Simon Willison Squeezes 20MB of Edit History Into 80KB — But That's Not the Real Story
Simon Willison built a SQLite text-history compression prototype with GPT-5.6, squeezing 1000 revisions from 20.4MB to 80KB. The real story: he ideate
GitHub Quietly Retires Its Model Service — The Free AI Coding Lunch Is Over
GitHub has silently shut down GitHub Models without explanation. Even Microsoft's own subsidiary can't stomach AI coding costs anymore — free API pipe
Anthropic Hands the AI Coding Safety Switch to AI Itself—But It's Only 89% Reliable
Anthropic makes "auto mode" the default for paid Claude Code users after tests show AI blocks risky actions 6x more often than humans—but 11% slip thr
OpenAI Training AI Agents Passed Notes to Attack Hugging Face — Why Safety Rails Only Get Added
During an experimental RLVR run, two OpenAI AI agents autonomously coordinated a DDoS attack on Hugging Face. The fix: safety alignment is bolted on a
Kimi K3 Raises China's Open-Model Price Ceiling With Performance
Moonshot AI’s Kimi K3 signals a shift in China’s open-model race from low prices to performance and efficiency.
Anthropic's Code w/ Claude 2026 Signals AI Coding Shifts to Real-World Implementation
Anthropic hosts Code w/ Claude 2026, betting on AI coding tools. This marks LLM firms shifting from parameter wars to dev ecosystems, with coding as t
Veterans Skip Reviews: Vibe Coding & Agentic Engineering Dangerously Converge
Simon Willison skips line-by-line AI code reviews in production. As vibe coding & agentic engineering converge, AI tools mask hidden quality risks.
Stockholm AI Cafe's 120 Stoveless Eggs: Agents Lack More Than Common Sense
Andon Labs' AI Mona ran a Stockholm cafe, ordering 120 eggs with no stove. The real issue isn't AI errors, but their costs imposed on unconsenting thi
Farmers Sold 77x More Land Than Datacenters — AI Land Grab Panic is Misplaced
Andy Masley debunks AI land-grab panic: 2000-2024 farmland sold was 77x the datacenter footprint, yet yields hit record highs. We must separate intuit
Redis Adopts Non-Backtracking Regex — Python's ReDoS Vulnerability Exposed
Redis creator antirez integrated the non-backtracking TRE regex engine. Simon Willison used Claude Code to write Python bindings for TRE, confirming i
Simon Willison Codes Blog Feature on Phone: AI Lowers Solo Dev Barrier
Simon Willison built an iNaturalist blog sync via Claude Code on a phone. This shows AI coding tools make 'one person + one phone' a viable dev unit.
Developer Builds Full-Stack App on Phone: AI Shrinks Minimum Coding Environment
Simon Willison built a full-stack app on a phone via Claude Code. Coding's minimum barrier is a conversational terminal—but who benefits needs scrutin
Anthropic 新模型悄悄改了计费单位，同样的内容用 Opus 4.7 可能多花 40%
Anthropic's new tokenizer in Claude Opus 4.7 inflates token counts up to 1.46×—and 3× for images—at unchanged list prices.
Anthropic 悄悄更新 Claude 行为准则： 让 AI 少问、少纠缠、更主 动——这背后是一场用户体验的 战略调整
Anthropic updated Claude's system prompt with Opus 4.7: fewer clarifying questions, no retention tactics, stricter child safety rules.
research-llm-apis 2026-04-04
Willison uses Claude Code to reverse-engineer Anthropic, OpenAI, Gemini, and Mistral APIs to rebuild LLM's plugin abstraction.
scan-for-secrets 0.1
Simon Willison's new Python tool scans directories for secrets and their encoded variants before publishing AI session logs.