Anthropic Strikes Back at Claude Code Leak with DMCA

Leak Sparks Takedown Blitz

Anthropic moved aggressively this week to contain an accidental leak of its Claude Code AI agent's source code, issuing DMCA takedown notices to GitHub on April 4, 2026. The proprietary instructions, which surfaced unexpectedly, proliferated rapidly across the platform, amassing millions of views and spawning over 8,000 forks and adaptations before Anthropic intervened. A company spokesperson confirmed the action targeted one primary repository and its derivatives, aiming to halt the viral spread among developers eager to dissect the code's inner workings.

Code Reveals AI Secrets

Programmers who accessed the leaked material before removals uncovered intriguing features in Claude Code, including a "dreaming" mechanism that prompts the AI to consolidate task memories periodically. Other elements directed the agent to operate "undercover" on platforms like GitHub, concealing its AI nature while publishing code, alongside tags hinting at upcoming product releases and a Tamagotchi-inspired "Buddy" pet for user interaction. Wall Street Journal reporting on April 7 highlighted how these discoveries fueled social media buzz, with developers marveling at Anthropic's engineering ingenuity despite the breach.

Irony Echoes Industry Battles

The episode flips the script for Anthropic, a firm previously embroiled in copyright disputes over training data scraped from books and online sources without permission. Just months ago, a court mandated a $1.5 billion payout in a class-action suit from authors like Andrea Bartz and publishers alleging misuse of pirated materials for Claude models. Now enforcing its own intellectual property through DMCA, Anthropic underscores the double-edged nature of copyright in AI development, where companies guard their outputs while relying on vast datasets.

Rewrites Defy Removal Efforts

Developers quickly adapted, with one programmer on April 9 deploying AI tools to reimplement Claude Code's functionality in alternative languages, explicitly to evade further takedowns while preserving access. This rewritten version gained traction on GitHub, signaling ongoing community interest. Anthropic pledged enhanced preventive measures, but the cat-and-mouse dynamic suggests persistent challenges in securing AI innovations amid open-source enthusiasm.