The Raptors Test the Fences

Muldoon explains that the raptors repeatedly probe different sections of the electric fence, remember the results, and search systematically for weaknesses.

Security tested only against normal behavior can fail against an adaptive adversary that learns from each unsuccessful probe. Robust containment and red teams must search for weaknesses the way the attacker would.

The raptors are biological attackers, not AI systems; their fence testing demonstrates the adversarial process robust defenses must anticipate, not a formal cooperative red-team exercise.