We consider a system with K processes of interest where for each process there is a source that generates status updates and send them to a monitor through a shared channel. We formulate a non-cooperative game in which each source is a player that chooses the generation rate of status updates of its process so as to minimize its cost, which is the weighted sum of the average Age of Information of its status updates and its load. We consider that the channel is a queue operating under the discipline LCFS with preemption in service. We show that, when the service rate of the status updates are equal, the Price of Anarchy of this game is upper-bounded by 2, whereas when the service rate are heterogeneous it is unbounded from above.