The adoption of AI tools in policing is increasing globally, with racial bias being a well-documented risk. However, representatives of affected communities are rarely included in decisions about AI adoption. We conducted a mixed-stakeholder deliberation workshop with 30 community representatives, police officers, and academics to assess the risks of 13 AI use cases in policing, with a focus on racial bias. The results showed that participants were broadly open to AI adoption, rejecting only three use cases, most notably recidivism risk assessment. Our analysis revealed that foregrounding racial equity did not narrow the deliberation, but instead led to a set of fundamental questions: does the tool work, will it deliver genuine benefit, and will that benefit extend to everyone? This integrated reasoning highlights the benefit of incorporating the racial bias lens into the risk-benefit analysis of AI use cases from the outset. Blogger's Review: This study explores the negotiation of AI risk boundaries in policing through a mixed-stakeholder deliberation workshop, finding that considering racial equity does not limit discussion but instead prompts deeper thinking about AI tool effectiveness and fairness, providing valuable insights for policing AI applications.