In a four-strategy Prisoner's Dilemma with peer punishers and rewarders, institutional rewards should target the enforcers, institutional punishment should target defectors only, and punishing enforcers destroys cooperation and welfare.
Regulatory Markets for AI Safety
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
abstract
We propose a new model for regulation to achieve AI safety: global regulatory markets. We first sketch the model in general terms and provide an overview of the costs and benefits of this approach. We then demonstrate how the model might work in practice: responding to the risk of adversarial attacks on AI models employed in commercial drones.
fields
cs.AI 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Co-evolution of social reward and punishment under institutional interventions
In a four-strategy Prisoner's Dilemma with peer punishers and rewarders, institutional rewards should target the enforcers, institutional punishment should target defectors only, and punishing enforcers destroys cooperation and welfare.