Simon Willison
AI Coding Agents Write Too Much, Making Code Reviews Harder
Claude Code and Codex produce exhaustive PR descriptions, but Simon Willison’s collapsible summaries expose a deeper problem: AI’s information burden.
Back to home
Simon Willison
30 articles tagged with this topic
Claude
Claude Pushes 3D Webpages Beyond Framework Limits as Code Generation Nears Reliability
Claude rewrote a Three.js 3D scene into native JavaScript and WebGL, showing AI can now deliver runnable projects, though human review remains essenti
Claude
Claude’s SQLite Extension Was Slower—AI Coding Still Needs Human Calibration
Claude built SQLite number-parsing extensions in C and Rust; after rigorous benchmarking, the revised C version outperformed Rust across key scenarios
Firefox
Firefox’s 24-Year Prefetch Waste Fixed by Datasette
Firefox prefetches Datasette’s “next page,” wasting server computation on pages few users open. Datasette disables the behavior.
Nilay Patel
AI Spending Is Outrunning Returns; Nilay Patel Says Products Still Aren’t Worth It
Nilay Patel says AI products lack visible returns while costs mount, making it unclear who can turn investment into value.
Paul Ford
Browsers Remain the Low-Cost Gateway to AI and Real-World Services
Paul Ford says browsers remain powerful AI gateways while their cost keeps falling, making safe, low-cost execution across real-world services a key c
Python 3.15
Python 3.15: Native Lazy Imports Fix AI Plugin Startup, Performance Flat
Python 3.15 ships PEP 810 lazy imports to fix slow AI plugin cold starts. Includes first built-in sampling profiler, but performance matches 3.14.
Dwarf Fortress
Dwarf Fortress Caves on Git After 20 Years — A Hardcore Dev's Compromise
Tarn Adams ran Dwarf Fortress solo for 20 years with no version control. In 2023 he finally relented — a rare look at principle vs. engineering pragma
Carson Gross
Carson Gross: Programming Isn't Typing Code — AI Won't Replace Engineers
Carson Gross: programming is problem-solving and complexity control, not typing code. We unpack this unsexy judgment and the risks it omits.
Ben Affleck
Ben Affleck Talks CNNs: AI Knowledge Spreads to Hollywood
Ben Affleck explains CNNs and tensors in plain English. Real story: AI knowledge spills from tech circles into Hollywood and mainstream discourse.
Simon Willison
Simon Willison: When AI Writes Your Blog, Human Voice Is the Moat
Simon Willison amplifies Michael Lynch's blogging checklist. As devs outsource writing to AI, "human voice" becomes scarce — readers grow hungrier and
Google
Google Open-Sources EmbeddingGemma 2: Hidden Costs of Closed-Source AI Search
Google releases EmbeddingGemma 2 open-source. Simon Willison warns closed-source embeddings create a vendor lock-in trap—Apache 2.0 is the exit.
Mistral
Mistral's Reasoning Model Lands in Open-Source Tool — Europe's Third Option
Mistral shipped Mistral Large Reasoning. Simon Willison's llm added same-day support (v0.16). Zero-day gap: European LLMs reach dev tools fast.
Simon Willison
LLM Price War Begins, Vulnapocalypse Hits — Willison: The Latter Is the Trouble
Simon Willison, one of the most-watched independent AI observers, flags two signals this month: the LLM API price war and an LLM tool "vulnapocalypse.
Anthropic
Claude 5.5: 30% Faster, 30% Cheaper, Now Free — Anthropic Poaches Users
Anthropic ships Claude Sonnet 5.5: 30% faster, 30% cheaper, same price. Mid-tier now powers free tier, beats ChatGPT free. Price war hits users.
Simon Willison
Letting AI Write Animations and Record Videos: A Parrot Party Chain Reaction
Willison shipped a pixel parrot animation + 15-sec video in 5 min via Claude Opus 5.5 + Claude Code + Playwright—no hand-code. The 'one-person team' j
Simon Willison
Simon Willison's Quiet Git Tool Update Reveals Daily Developer Friction
Simon Willison bumped commit-rewriter to 0.2, adding support for rewriting commits on non-default branches. Small upgrade, but it exposes an overlooke
Paul Dix
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
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.
Simon Willison
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
Jeremy Morrell
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
Qwen
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.
Simon Willison
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
Simon Willison
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.
OpenClaw
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
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
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
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
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
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