Open-source security in the AI exploit era: Patch without changing a thing

AI models like Anthropic's Mythos and OpenAI's GPT-5.5 are rapidly discovering thousands of new software vulnerabilities, drastically shortening the time from vulnerability disclosure to exploitation. Open-source software, heavily relied upon in modern development, faces intense pressure as many vulnerabilities remain unpatched due to volunteer maintainer constraints, prompting urgent calls for new security strategies.