Anthropic Accidentally Leaks Claude Source Code

Anthropic has inadvertently disclosed the instructions behind its Claude Code AI agent. The exposure could provide competitors with strategic insight into how the model is created and could introduce potential security risks.

The leak did not compromise customer data or the core mathematical frameworks of its AI models, a spokesperson for Anthropic told the WSJ. The incident was attributed to a packaging error rather than a breach of security.

However, the disclosure of Anthropic's proprietary methods and tools that help Claude work as a coding agent, also known as a harness, presents a risk of being replicated by competitors without the need for reverse engineering.

The company, valued at $380 billion, is experiencing increased usage of its Claude Code and is considering a public offering later this year. In February, Anthropic announced that it had raised $30 billion in Series G funding led by GIC, D.E. Shaw Ventures, Coatue, among others.

Last month, San Francisco Federal Court District Judge Rita Lin sided with Anthropic in its request for a preliminary injunction in its legal battle against the Trump administration, calling it "illegal First Amendment retaliation."

This decision temporarily halts the government's actions to blacklist the AI company and prevents the enforcement of a directive from President Donald Trump that bans federal agencies from using Anthropic's Claude models.

Photo Courtesy: Koshiro K on Shutterstock.com

This content was partially produced with the help of AI tools and was reviewed and published by Benzinga editors.

Market News and Data brought to you by Benzinga APIs

To add Benzinga News as your preferred source on Google, click here.