This paper identifies three categories of safety risks posed by allowing multiple users to engage with small Unmanned Aerial Systems (sUAS) and offers five recommendations on how to reduce or mitigate these vulnerabilities. Data from sUAS can benefit multiple experts at a disaster who may not be familiar with robots or colocated with the pilot. Two different styles of interfaces have been developed and tested with responders conducting exercises to facilitate team coordination with a quadrotor at Texas A&M Engineering Extension Service's Disaster City® over a four year period. The two interfaces illustrate three distinct categories of safety concerns: unsafe control regimes, loss of situation awareness, and increased stress. Five recommendations are proposed to mitigate or eliminate the safety concerns: separate the payload camera from the platform, giving the pilot a dedicated "pilot-cam" and the experts a fully gimbaled payload; use artificial intelligence to resolve conflicts between competing directives from multiple experts; allow the pilot, or a software agent, to turn off the expert's ability to control or communication; use multi-modal warnings rather than rely on visual cues; and add guarded motion to prevent collisions.