'The Irony Is Rich': Anthropic Issues Copyright Takedowns to Stem Claude Code Leak

Anthropic is working to mitigate damage from a massive leak of its Claude Code programming assistant, which exposed over 500,000 lines of code earlier this week. The leak, which revealed inner workings of the tool, did not include private user data, the company confirmed.

The AI firm has issued copyright takedown notices to platforms like GitHub, where thousands of copies of the leaked code have spread. A spokesperson told The Wall Street Journal the leak was due to a human error in release packaging—not a security breach—and that Anthropic is rolling out preventive measures.

Online communities like Reddit have highlighted the irony of Anthropic’s actions. Last year, the company paid $1.5 billion (£1.14 billion / AU$2.18 billion) to authors whose books were pirated and used as training data for its AI models. Reddit users reacted with comments like "the irony is rich" and "good luck with that," with some labeling AI bots as "plagiarism machines."

Claude Code is not the consumer-facing Claude chatbot but a developer-focused programming assistant widely regarded as one of the industry’s best. This makes the leak particularly impactful, as experts have raised concerns about AI-generated code lacking security and safety best practices.

Recent days have also seen usage limits restricted for Claude users—even paying customers—reflecting the balance AI companies face between user access and the high costs of running AI systems.

Sources

• TechRadar: "The irony is rich': Anthropic issues copyright takedown requests in attempt to stem Claude Code leak"
• The Wall Street Journal (as reported in the article)