Politics

AI rewrite of leaked code sparks legal debate on copyright boundaries

Thursday, 23 April 2026 12:00AM UTC

The accidental leak of Anthropic’s Claude code has ignited a debate on how AI-generated rewrites impact copyright laws, challenging traditional notions of authorship amid rapidly evolving digital tools.

The accidental exposure of Anthropic’s Claude Code source code has become more than a security embarrassment for the company; it has also turned into a live test of how copyright law functions when AI can rapidly transform, rework and redistribute software. The New York Times reported that Sigrid Jin, a University of British Columbia student, copied the leaked material and then used AI assistants to rewrite it in another programming language before posting that version online, prompting renewed debate over whether a machine-assisted rewrite can escape claims of infringement.

That dispute sits within a wider legal fog around software, licensing and AI-generated code. Ars Technica recently examined how an AI-produced rewrite of the chardet library raised questions over whether an apparently fresh version can still amount to a "clean-room" copy, especially when the structure and behaviour of the original work remain recognisable. Legal commentators have also warned that AI tools are making it harder for developers and companies to rely on older assumptions about authorship, ownership and copyright protection for code.

Anthropic has tried to contain the fallout with a large-scale takedown effort. According to reporting from AI Coin and TechFlow Post, the company issued more than 8,000 copyright removal requests after the leak spread across the internet, with thousands of posts pulled from GitHub. But the episode has also highlighted the limits of enforcement on decentralised platforms, particularly when copies can be reposted, transformed or shared faster than rights holders can respond.

For Jin, the point was not simply to avoid a legal complaint. The New York Times said he viewed the rewritten code as a distinct work and used the episode to make a broader argument about what copyright means in an era when AI can reproduce creative output at speed. That tension between legal doctrine and technical possibility is now at the heart of the Anthropic case: if software can be rewritten almost instantly by machines, the law may still apply, but it is being asked to police a far more fluid and evasive form of copying than it was designed for.

Source Reference Map

Inspired by headline at: ^[1]

Sources by paragraph:

Paragraph 1: ^[2], ^[4]
Paragraph 2: ^[2], ^[5], ^[6]
Paragraph 3: ^[3], ^[7]
Paragraph 4: ^[1], ^[5], ^[6]

Source: Noah Wire Services

More on this

https://www.vogelitlawblog.com/2026/04/does-the-copyright-act-apply-to-ai-anymore/ - Please view link - unable to able to access data
https://arstechnica.com/ai/2026/03/ai-can-rewrite-open-source-code-but-can-it-rewrite-the-license-too/ - This article discusses the legal complexities arising when AI tools are used to rewrite open-source code. It examines a case where an AI-generated version of the chardet library was released under a different license, leading to debates over whether such AI-generated code can be considered a 'clean-room' rewrite and how it affects copyright and licensing laws.
https://www.aicoin.com/en/article/524907 - This piece explores the aftermath of Anthropic's accidental source code leak, highlighting the company's issuance of over 8,000 copyright takedown requests. It delves into the challenges of enforcing copyright on decentralized platforms and the broader implications for AI companies' 'safety-first' reputations.
https://inshorts.com/amp_news/techie-rewrites-claude-codes-leaked-source-code-with-rival-ai-over-copyright-issues-1775023254014 - This article reports on South Korean developer Sigrid Jin's use of OpenAI's Codex to rewrite the leaked source code of Anthropic's Claude Code. Jin's approach aims to circumvent potential copyright issues by creating a version that captures the original's architectural patterns without copying proprietary source code.
https://www.techdirt.com/2026/04/03/can-agentic-ai-coding-tools-finally-end-copyright-for-software-while-re-inventing-open-source/ - This article examines the impact of generative AI systems on software development and copyright law. It discusses how AI tools are increasingly used to write code, potentially challenging traditional notions of authorship and ownership in open-source software.
https://www.carltonfields.com/insights/publications/2026/ai-makes-securing-copyright-protection-for-software-code-tricky-bloomberg-law - This publication addresses the challenges in securing copyright protection for software code in the era of AI. It highlights how the shift from human-written to AI-generated code complicates traditional copyright strategies and discusses the implications for companies and developers.
https://www.techflowpost.com/en-US/article/30966 - This article details the events following Anthropic's accidental exposure of Claude Code's source code, including the company's issuance of over 8,000 copyright takedown requests. It also covers the developer community's response and the broader implications for Anthropic's 'security-first' image.

Noah Fact Check Pro

The draft above was created using the information available at the time the story first emerged. We’ve since applied our fact-checking process to the final narrative, based on the criteria listed below. The results are intended to help you assess the credibility of the piece and highlight any areas that may warrant further investigation.

Freshness check

Score: 8

Notes: The article was published on April 22, 2026, which is recent. However, the content references events from March 31, 2026, and April 1, 2026, indicating that the information is not entirely fresh.

Quotes check

Score: 7

Notes: The article includes direct quotes attributed to Sigrid Jin and Meaghan Tobin. However, the exact wording of these quotes cannot be independently verified through the provided sources.

Source reliability

Score: 6

Notes: The article is published on the Vogel IT Law Blog, which appears to be a niche legal blog. While it may have expertise in the field, its reach and reputation are limited compared to major news organisations.

Plausibility check

Score: 7

Notes: The events described, including the accidental leak of Anthropic's Claude Code source code and the subsequent actions by Sigrid Jin, are plausible and align with known incidents. However, the article's reliance on a single source for these events raises concerns about the completeness and accuracy of the information.

Overall assessment

Verdict (FAIL, OPEN, PASS): FAIL

Confidence (LOW, MEDIUM, HIGH): MEDIUM

Summary: The article presents information on recent events related to AI and copyright law, referencing a New York Times article. However, the reliance on a single source, the inability to independently verify direct quotes, and the limited reach of the publishing platform raise significant concerns about the article's reliability and completeness.

AI and copyright
software law
leaked source code