When AI Moves Fast and Breaks Copyright: The Anthropic Case

Anthropic’s recent code leak is more than a technical slip, it’s a window into a deeper problem in the AI industry. After accidentally exposing internal code through a leftover file, the company tried to scrub over 8,000 instances from the internet, only to later admit just 96 were legitimate targets. That overreach, combined with a prior $1.5 billion copyright settlement and opaque data practices like Project Panama, paints a picture of an industry moving faster than its own accountability systems. At its core, this isn’t just about one leak, it’s about who owns the data powering AI, and whether the companies building it truly understand or respect those boundaries.