Blog
Notes and analysis on AI development.
MiniMax M3 puts 1M-token open-weight agents to a cost test
MiniMax M3 arrives with a 1M-token context window, agentic coding support, and an open-weight promise. The launch separates API access from verification work.
The White House AI security order puts frontier models on a 30-day review track
The White House AI security order ties CISA guidance, a Treasury-led vulnerability clearinghouse, classified cyber benchmarks, and pre-release frontier model access into 30-day and 60-day deadlines.
Adafruit received a Flux.ai demand letter, and AI PCB tools need verifiable provenance
Adafruit published a Flux.ai demand letter dispute. The real builder question is how AI PCB tools prove provenance, data boundaries, and human review.
Codex reaches 5 million weekly users as knowledge work moves into agents
OpenAI says Codex now has more than 5 million weekly users, with knowledge workers using it for reports, contracts, spreadsheets, data analysis, and internal tools.
Claude -p automation gets a $20 credit line on June 15
Anthropic is moving Claude Agent SDK and claude -p usage out of subscription limits. Here is what changes for CI, scripts, and third-party agent apps.
Stanford CS336 publishes CLAUDE.md, drawing a sharper line for AI coding assignments
Stanford CS336 uses a repo-level CLAUDE.md to keep coding agents in a teaching-assistant role, allowing concept help while banning solution code.
ChatGPT Sheets flaw exposed 12 workbooks even with approvals disabled
PromptArmor disclosed an indirect prompt injection chain in ChatGPT for Google Sheets. OpenAI responded by removing Apps Script generation.
Perplexity Search as Code Runs Agent Search in Python
Perplexity introduced Search as Code, a Python sandbox architecture for turning repeated agent retrieval into executable search pipelines.
Claude Code study tracks 5,838 developers, +41 monthly commits, and +0.83 languages
An arXiv paper analyzes GitHub activity before and after Claude Code adoption across 5,838 developers, with commit, repository, language, and causality caveats.
Anthropic Files Draft S-1, Putting Claude Code Costs Under Review
Anthropic has confidentially filed a draft S-1 with the SEC. Claude Code growth, compute costs, and enterprise concentration are moving toward public-market scrutiny.
Copilot Auto can now route individual plans to evaluation models
GitHub Copilot Auto can serve evaluation models to individual plans. Opt-out controls, model visibility, and security prompts are now operational checks.
Alphabet raises $80B for AI compute as Gemini demand becomes a balance sheet problem
Alphabet announced $80B in equity offerings for AI compute expansion. The financing shows what Gemini APIs, agents, and Google Cloud capacity now cost.