Agents, some with a bias, decide between undertaking a risky project and a safe alternative based on information about the project’s efficiency. Only a part of that information is verifiable. Unbiased agents want to undertake only efficient projects, but biased agents want to undertake any project. If the project causes harm, a court examines the verifiable information, forms a belief about the agent’s type, and decides the punishment. Tension arises between deterring inefficient projects and a chilling effect on using the unverifiable information. Improving the unverifiable information always increases overall efficiency, but improving the verifiable information may reduce efficiency.