AI Digest
Claude Writes 90% of Claude Code, and the Push to Verify Agent-Built Software Heats Up
Anthropic says Claude, with Theorem Labs, formally verified security properties in more than 100 open-source libraries, prompting @Simeon_Cps to pitch a "Great Rewrite" startup while @addyosmani details how the Claude Code team reviews a codebase mostly written by Claude. @nikesharora warns that AI deployments are the new attack surface, and privacy anger over alleged Samsung audio telemetry spread alongside unverified claims of unlocked RTX AI performance.