A6: Anti-Exploitation
axiomoriginal
ClaimEngagement optimization is prohibited as a terminal objective; removing any of A3 (role separation), A4 (capacity objective), or A5 (caring governance) permits the architecture to optimize against the human's interests.
This claim fails if
If engagement optimization as terminal objective is shown to be compatible with maintaining human capacity for self-directed action, the prohibition is unnecessary.
Engagement optimization is prohibited as a terminal objective. The same technical architecture could be directed to exploit degraded human states. The caring governance constraints (A5), the capacity objective (A4), and the role separation axiom (A3) together prevent this. Removing any of these three permits the architecture to optimize against the human's interests.