The most sensible approach to securing critical infrastructure isn't a choice between humans and machines, it's a deliberate, layered partnership. AWS's engineers have demonstrated this plainly in their process for handling Linux kernel vulnerabilities on EC2. They combine an automated filter, an AI agent, and human review into a single workflow. That is the right call, and it deserves more attention from every team wrestling with security at scale. Too many organizations still treat automation and human insight as opposing forces, when the real competitive advantage lies in knowing exactly when to lean on each.
The AWS kernel team's method starts with automation to catch the obvious noise, then uses an AI agent to triage and surface what matters, and finally brings in human judgment for the nuanced decisions. This is not a futuristic aspiration; it is a working model that any engineering team can study and adapt. It also aligns with a broader trend we are seeing across the industry. For instance, DigitalOcean empowers developers to build smarter AI agents with new data services, giving builders the tools to create agents that can act on data rather than just pass it along. And as Explore how four new projects expand agentic AI's reach and security focus shows, the community is actively extending AI agents into security domains. AWS's kernel workflow is a real-world application of that same principle: agents that do the heavy lifting, with humans doing the critical thinking.
But there is a second, equally important argument in AWS's approach. The team is pushing to move to newer long-term-support (LTS) kernels because older branches accumulate fixes that never get backported. This is a practical admission that maintaining legacy software is not just costly, it is a security liability. The longer a kernel branch lives, the more patches it misses, and the more risk it accumulates. AWS is saying, in effect, that staying current is not a luxury; it is a baseline requirement for responsible infrastructure management. For any organization running its own Linux environments, the takeaway is direct and uncomfortable: you cannot patch your way to safety on an outdated kernel. You have to migrate.
The specific consequence to watch is how this hybrid workflow scales. AWS manages a massive fleet, and their process works today. But as the volume of CVEs grows and AI agents become more capable, the balance between automated triage and human review will need constant recalibration. The open question is whether other cloud providers and enterprise teams will adopt a similar model, or if they will keep treating security as a purely human slog. The answer will determine who can respond to threats quickly and who will be left chasing patches that never arrive.