The fix increases the temperature from 1000 to 1200 in the first run 10000. Then, since you did not specify otherwise, another run is executed where the temperature is again raised from 1000 to 1200 as the region moves. As the simulation proceeds, the thermo output will likely report the global temperature computed on all atoms, which includes the sample’s part that is not thermostated but will likely absorbs energy from the laser_region. Show us the temperature profile of your simulation, and tell us what is unexpected about your findings.
Update Just noticed that you posted the same question on a new thread. As pointed out by @baerb
So for the region to behave any differently, the Langevin thermostat needs to be applied only to the atoms in the specific region:
fix laser_heating laser_region langevin 1000 1200 400 654759