Merging Ai and Security
Cybersecurity Engineer building secure AI systems, agentic applications, and developer tools. Passionate about AI security, cloud, and turning research into production-ready software. Sharing projects, technical deep dives, and lessons learned along the way.
Markdown supported · ⌘↵ to post
AI has definitely shifted the bottleneck from writing code to evaluating decisions.
One thing I'd add is that AI-generated code often passes code review because it looks clean. The real review should focus on architecture, failure modes, security boundaries, and whether the implementation actually matches the business intent.
I've also noticed teams getting better results when AI is treated as a fast implementation partner rather than an autonomous engineer. The quality of the system still comes down to the questions humans ask, the constraints they define, and the production standards they enforce.
The responsibility never moved from engineers it just moved higher up the stack.
Thanks, Mateo! That's a great perspective. I especially agree that the bottleneck has shifted from writing code to evaluating architecture, trust boundaries, and security assumptions. That's one of the motivations behind Bulwark. AI can generate or assemble systems incredibly quickly, but someone still needs to verify what is being integrated, why it's trusted, and whether the resulting agent has capabilities beyond what it actually needs. AI can accelerate implementation, but accountability and security architecture remain human responsibilities.
Kartik N V J K
The "harmless parts, dangerous combination" point (read-files tool plus send-email tool equals exfiltration path) is the risk I think teams underrate the most, because each capability passes review in isolation. The rename-trick detail is a good reminder that inspecting by file extension is theater; content inspection of the serialized model is the only thing that holds. I would be curious whether you scope tool combinations at the policy layer too, since even a fully inspected model can be talked into chaining two safe tools by an injected instruction in retrieved content.