This year, nearly 50,000 security issues have been identified – around 130 every day. Traditional approaches based on manual assessment and backporting are no longer sufficient to keep pace with this reality. New approaches are needed, including the use of AI agents and new products capable of addressing these challenges more effectively. One possible solution is the use of minimal containers built on fully automated agentic systems designed to minimise the number of vulnerabilities. Automation alone, however, is not enough. Red Hat’s Lightwell project shows how the industry is moving from reaction to prevention. The USD 5 billion initiative brings together 20,000 engineers and AI agents to validate vulnerabilities, backport fixes to specific dependency versions, and contribute patches directly to upstream projects. The session will explore how AI agents can assess whether vulnerabilities are genuinely exploitable, port code to safer programming languages, and fundamentally transform open-source security in an era of rapidly growing vulnerability volumes.