Anthropic’s Claude Code, an agentic AI that drafts entire modules, hit headlines after a March 31, 2026 update accidentally exposed 512,000 lines of its source on GitHub. The repository garnered 100,000 stars in a day, then faced DMCA takedowns.

The incident forced developers to confront three ownership pillars: meaningful human authorship, employment‑based work‑for‑hire doctrines, and lurking GPL footprints from training data. Courts confirmed that purely AI‑generated code lacks copyright protection, leaving it effectively public domain.

For coders, the rule of thumb is simple: document every architectural choice and test run, keep side projects on personal hardware, and avoid employer‑licensed tools for private work. Even a single line derived from GPL‑licensed material can trigger copyleft obligations.

Ultimately, the Claude Code episode underscores that AI‑assisted code requires deliberate human oversight to secure legal custody and avoid accidental open‑source contamination.