Anthropic on Tuesday said it unintentionally released a portion of the internal source code for its AI coding assistant, Claude Code, due to what it described as a “human error”.
According to the company, an internal-use file was mistakenly included in a software update, which linked to an archive containing nearly 2,000 files and around 500,000 lines of code. The material was quickly copied and shared on developer platform GitHub.
“Earlier today, a Claude Code release included some internal source code. No sensitive customer data or credentials were involved or exposed,” an Anthropic spokesperson said.
The company clarified that the incident was caused by a release packaging mistake and was not the result of a security breach.
A post on X containing a link to the leaked code had garnered more than 29 million views by early Wednesday.
The leaked material reportedly relates to the internal structure of the tool and does not include confidential data tied to Claude, Anthropic’s core AI model.
Parts of Claude Code’s source code were already partially known, as independent developers had previously reverse-engineered the tool. An earlier version of the assistant’s code was also exposed in February 2025.