Code on computer screen

Anthropic Scrambles to Contain Major Leak of ‘Claude Code’ Secrets

Anthropic’s Claude Code AI faced exposure online, prompting swift takedown efforts to safeguard proprietary details.

Anthropic is trying to contain the fallout after internal code tied to its Claude Code AI agent was exposed online, giving developers and rivals a closer look at how the company’s coding assistant works, according to The Wall Street Journal. Anthropic used a copyright takedown request to remove thousands of copies and adaptations from GitHub before narrowing that request to 96 repositories, the Journal reported.

Anthropic said the incident did not expose customer data or the model weights behind its flagship AI systems. The company described the problem as a packaging mistake caused by human error, not a security breach. Still, the leaked material appears to have included commercially sensitive instructions, tools and techniques used to turn Claude into a coding agent.

A GitHub takedown notice suggests the code has been pulled from over 8.1 thousand repositories. The message reads: “Because the reported network that contained the allegedly infringing content was larger than one hundred (100) repositories, and the submitter alleged that all or most of the forks were infringing to the same extent as the parent repository, GitHub processed the takedown notice against the entire network of 8.1K repositories, inclusive of the parent repository.”

Still, a post on X with a link to Anthropic’s code had more than 33.3 million views as of 4:00 p.m. ET.

That matters because the race to build AI coding tools is moving fast, and much of the edge comes not just from the model itself, but from the surrounding “tooling” that helps it perform useful work for developers. The Journal said the leak offered outsiders a road map for copying Claude Code features without having to reverse engineer them first.

The leak also created a second problem. By exposing more of Claude Code’s internal logic, it may give security researchers and threat actors new material to study for weaknesses, even if one outside expert told the Journal the incident looked more embarrassing than dangerous. The same report said programmers quickly began dissecting the code, spotting hidden features, future product hints and design choices that helped explain why Claude Code has gained traction with developers.

Image by StockSnap from Pixabay

Total
0
Shares

Leave a Reply

Previous Article
Beautiful sculptures and faces

Google Vertex AI Flaw Turn Agents into ‘Double Agents’

Next Article
EU Flags in front of building

EU Supply-Chain Cloud Breach Tied to Trivy Compromise

Related Posts

Discover more from Security Point Break

Subscribe now to keep reading and get access to the full archive.

Continue reading