We study a situation in which a regulator relies on risk models that banks produce in order to regulate them. A bank can generate more than one model and choose which models to reveal to the regulator. The regulator can find out the other models by monitoring the bank, but in equilibrium, monitoring induces the bank to produce less information. We show that a high level of monitoring is desirable when the bank's private gain from producing more information is either sufficiently high or sufficiently low (e.g., when the bank has a very little or a very large amount of debt). When public models are more precise, banks produce more information, but the regulator may end up monitoring more.