2026-10-12 · ai, engineering
Rogue AI agents and accountability
Calling it a rogue AI agent doesn't remove the responsibility to ship with guardrails.
“Rogue AI agent” is becoming a popular term. I think it's just a short way to say:
I pushed bad code without guardrails which broke something, but of course it's not my fault.
In the old days you could get fired for this attitude. Now it's just a way to shift blame while attracting even more investor attention.
Growth hack: next time someone's AI slop breaks prod, make sure to publish a company PR on how AI tried to destroy your startup.
Originally posted on LinkedIn.