There is an equation governing fatality along the lines of:

D% = A% * F%

A% = Chance of getting into an accident

F% = Chance of a given accident being fatal

D% = Chance of cyclist dying in a given time period

There is a possibility that the increase in A% caused by wearing a helmet is greater than the corresponding decrease in F%, in this case it is indeed safer not to wear helmets. To justify this there would have to be some VERY conclusive statistical evidence support both sides of the calculation, or the company will open itself up to a massive lawsuit the first time someone has an accident.