Anthropic released Claude Sonnet 5, a new AI model with performance close to Opus 4.8 at lower prices, and updated its API by removing temperature, top_p, and top_k parameters. The model's tokenizer i…
A developer created a single-page React app called 'The AI Compass' that categorizes users into AI archetypes like 'The Garage Tinkerer' based on a quiz. The app uses a script trick to avoid a build s…
Simon Willison released shot-scraper 1.10 with a new `video` command that records Playwright-driven video demos of web applications from a storyboard YAML file, enabling coding agents to produce visua…
Shot-scraper 1.10, a CLI tool for website screenshots and scraping, was released on June 30, 2026, introducing a new video storyboard feature that allows users to record video demos of their work.…
A developer shared a tiny AppleScript tip to count the number of open Safari tabs using the command 'osascript -e "tell application \"Safari\" to count tabs of every window"'.…
DeepReinforce released Ornith-1.0, a family of self-scaffolding LLMs for agentic coding built on Gemma 4 and Qwen 3.5, achieving state-of-the-art performance among open-source models of comparable siz…
Jon Udell argues against the phrase 'human in the loop,' advocating instead for a model where humans invite AI agents into their existing workflows as team members, not as autonomous decision-makers. …
Dean W. Ball warns that delays in releasing frontier AI models are eroding the narrow window for labs to recoup enormous training costs, and that the ongoing AI infrastructure buildout assumes a globa…
Timothy B. Lee criticized the notion that large language models require no skill or learning curve, comparing it to the false assumption that managing employees is effortless because they follow order…
After 2,000 people attempted to hack an AI assistant over 6,000 times, spending $500 in tokens and triggering a Google account suspension, no one succeeded in leaking a secret. The assistant, powered …
Two AI review agents from competing vendors entered a disagreement loop over a pull request, generating 340 comments and $41,255 in inference spend before Finance revoked both API keys. One vendor's m…
OpenAI announced a limited preview of the GPT-5.6 series, including flagship model Sol, balanced model Terra, and fast/affordable model Luna. Terra matches GPT-5.5 performance at half the cost, while …
Datasette released version 0.3a2 of datasette-export-database, fixing a critical compatibility issue where pyproject.toml incorrectly pinned the dependency to datasette==1.0a27, making the plugin unus…
Simon Willison created a GitHub repository that uses AI-generated scripts and GitHub Actions to build a 66MB SQLite database from Mozilla's browser compatibility data, hosted with open CORS headers fo…
Tom MacWright reports seeing job applications that are clearly cowritten by an LLM, linking to LLM-generated portfolios and GitHub projects with LLM-generated commit messages. He argues that such appl…
Simon Willison built a test harness combining OPFS and Pyodide to explore whether Datasette Lite can edit persistent SQLite files stored in the browser's Origin Private File System. The tool allows te…
Researchers Charles Ye, Jasmine Cui, and Dylan Hadfield-Menell found that large language models suffer from 'role confusion,' mistaking the style of text for its actual content, leading to successful …
Developer Simon Willison ported the Moebius 0.2B image inpainting model to run in a browser using WebGPU, bypassing the original requirement for PyTorch and NVIDIA CUDA. The project, completed with as…
Sqlite-utils 4.0rc1 has been released, introducing migrations and nested transactions to the SQLite database utility tool. The release candidate adds new features for managing database schema changes …
Sean Lynch argues that the Model Context Protocol's key advantage over skills and CLI is isolating authentication outside the agent's context window, potentially reducing MCP to a pure auth gateway fo…