Anthropic Claude AI source code leak: ‘Human error’ sparks security concerns

Anthropic is once again in headlines due to accidental release of its Claude AI agent source code.

The recent exposure is a second mistake made by Anthropic in a week. Earlier, the US-based company accidentally revealed its upcoming “Claude Mythos” due to a content management error on March 27, 2026.

A post on X sharing a link to the leaked code was viewed by more than 30 million people on Wednesday.

Soon after the release of confidential codes, an Anthropic spokesperson issued a statement, “Earlier today, a Claude Code release included some internal source code. No sensitive customer data or credentials were involved or exposed. “

“This was a release packaging issue caused by human error, not a security breach. We’re rolling out measures to prevent this from happening again,” the statement read.

The leaked internal-use file included an archive containing nearly 2,000 files and 500,000 lines of code, which were quickly copied to developer platform GitHub. However, fortunately, no confidential data from Claude were found.

Although the recent exposure of code did not result in any kind of data breach, the grave incident raises major concerns surrounding Anthropic operational security.

While independent developers had previously reverse-engineered portions of Claude Code, this incident follows a prior exposure of the assistant’s source code in February 2025.