Anthropic has been in the spotlight for its commitment to responsible AI development. However, two recent mishaps have brought its human fallibility into stark relief.

Last week, nearly 3,000 internal documents were mistakenly made public, including a draft blog post about their new model. Today, they released version 2.1.88 of Claude Code, only to find that it exposed over 512,000 lines of source code.

While Anthropic has stated that these issues are human errors and not breaches, the stakes could not be higher. Their command-line tool is a formidable competitor in the AI space, influencing rivals like OpenAI's decision to pivot away from video generation products.

The leaked code provides valuable insights into how Claude Code operates, but whether this will impact the company’s future remains to be seen. For now, one can only wonder who might have had a quiet panic session over their job security after these events.