This blog post details the experiences of the curl maintainers with the OSS-Scanner service by Anthropic, which identified 45 security vulnerabilities in curl. It discusses the severity of the findings, the nature of the vulnerabilities versus mere bugs, and the quality of AI-generated reports compared to human insights. The author also reflects on future participation in such programs due to the trivial nature of some findings and encourages community participation in reviewing and addressing reported issues.