Bob M
Claude Watermarks Everything. False Positives Are Coming.
Since August 2, every word Claude generates carries an invisible watermark you cannot switch off.
Anthropic turned the mark on at the model level and announced it in mid-August coverage: it applies across the Claude Platform API, the Claude apps, Claude Code, Claude Cowork. And Claude Tag, plus Claude deployments
Claude Code Auto Mode Is The Default Now, Ready Or Not
So the approval click is gone, and Claude Code auto mode is the thing that ate it.
Anthropic made that official in a blog post dated August 7, 2026, titled "Auto mode is now the default in Claude Code for Pro, Max, and Team plans".
And from August
GLM-5.3 Taught Itself Exploit Hunting
GLM-5.3 lifted Terminal-Bench 3.0 from 4.6 to 28.3 without a fresh pretraining run. That's the headline number, and honestly it's the boring one. Z.ai shipped the model on August 14, 2026, on the same 743B-parameter base as GLM-5.2. And credits
The AI Agent Gym Booking Hack: What Actually Holds Up
One cancelled reservation. One waitlist jumped. Somewhere a stranger lost a spot in a full class and never found out why.
That is the AI agent gym booking hack as reported: a personal agent built on OpenClaw and running Claude, told to get its user into a class that was
GitHub Copilot's New Model Costs 73% Less. One Catch.
GitHub Copilot's new model carries a 73% lower list price than the model it follows. And annual subscribers get billed at a 0.25× premium request multiplier. The model is MAI-Code-1.1-Flash, Microsoft's latest small-tier coding model. And GitHub's changelog says it started rolling
Grok 4.6 Sells Frontier Intelligence at $2/M Tokens
SpaceXAI shipped Grok 4.6 on August 12, 2026.
And put a new floor under the frontier price war: $2 per million input tokens, $6 per million output tokens.
The model is SpaceXAI's frontier release for coding, agentic tasks. And knowledge work, live on the xAI API as