It sounds like we can train other agents to procedurally and adversarially generate these honeypots. Then it would become an evolutionary deception-detection arms race and also eliminate the concern of honeypots being “discovered,” since there are a boundless number of other honeypots being proliferated at any given point in time. Thoughts?
It sounds like we can train other agents to procedurally and adversarially generate these honeypots. Then it would become an evolutionary deception-detection arms race and also eliminate the concern of honeypots being “discovered,” since there are a boundless number of other honeypots being proliferated at any given point in time. Thoughts?