Anthropic's AI coding assistant, Claude Code, was discovered in a publicly accessible database containing details for features yet to be announced. The company, which claims to have zero control over when and how such information is made public, has reportedly removed over 8,000 copies of the Claude Code source code from its repository. The leak, which has been described as human error, also gives hackers and malicious actors access to the source code. Anthropic's competitors may be able to copy some of Claude's functionality without reverse engineering the underlying code. The firm's plans for a new model called Mythos were discovered in an accessible database last month.