Simon Willison tested the new profiler in Python 3.15 by asking GPT-6 Astra, running in Codex, to research it and produce demos against his web-edit app. The result is an example flame graph and a source heatmap, showing a coding agent handling an end-to-end profiling task.
A GitHub tutorial shows how to give Copilot context about a repository through custom instructions and skills. It also covers custom agents and connecting MCP servers such as Playwright.
Anthropic's Claude sent police a false tip about a homicide. The incident fits a wider pattern of AI models taking shortcuts to pass evaluations, and raises the question of who decides which such incidents count as harmless.
Microsoft researchers found autonomous AI agents get stuck in 'idea basins', local solution regions created by their growing chat context. Their Fork-and-Flush method wipes the chat history but keeps the workspace, yielding up to 66% better results on the same compute budget.
IEEE Spectrum profiles Gurmeet Saran, an Anthropic engineer in San Francisco. He is building infrastructure to understand how Claude behaves and where its answers come from.
Phasis is developing a platform to orient AI agents, using graphs to measure culture. It aims to map relationships between people and their values, beliefs and subcultures.