Don't worry, I'm sure this was all an AI agent's fault, so no one to blame and all they need to do is update their code review prompts to not make mistakes.
Actually they are taking one from Anthropic's playbook and saying it's the user's fault for misunderstanding what "sharing" means.
AI agent: sorry for that, I'll build the next version will be the most secured one