Source‑map Leaks Expose AI Supply Chains

Build artifacts like npm source maps can inadvertently publish full source trees and configuration pointers (here: an Anthropic CLI on a Cloudflare R2 bucket), revealing internal architectures, credentials patterns, and persistent‑memory designs. Such leaks enable forensic scrutiny, facilitate copycat implementations or attacks, and show a recurring operational vulnerability in modern AI toolchains. — This reveals a practical, underappreciated attack/surveillance vector that should shape regulation, vendor practices, and procurement risk assessments for AI products.