The hazard of these cars is surprise: They can be operated with dynamics that surrounding traffic do not or cannot expect.

Speed differentials are hazardous because of limits of human attention.

Consider the simple case of an urban crosswalk: The rate of vehicles vs. pedestrians has an attention hazard for both: there's a speed beyond which the pedestrian can not safely judge a crossing because the vehicle will pass in a duration shorter than the attention span of the ped. Similarly there's a speed at which the operator can not react to appearance of ped.

Simple V differences are a well known hazard of surprise. EVs create another hazard which arise from relatively high delta-V dynamics. The vehicles can accelerate at surprising rates and therefore appear in places that operators don't expect them to be. For example, merging requires anticipation that the surveyed gap will be maintained during the crossing. Surrounding vehicles operating with disproportionate V or high dV upset the balance of the gap.

Surprise can have cascading effects, where surprised operators lose attention which in turn causes further hazards.

We might suspect that given the surprising power of EVs, their saving grace is that their operation in urban traffic tends towards operator passivity. But this tendency obviates power: the proper power level is that which permits the vehicle to operate fluidly with traffic.

Regardless of attention spans for operators, the simple mechanical comfort of occupants places very low limits on vehicle dynamics.

Vehicle power was already becoming well balanced before EVs, and EV development should be further refining the balance in favor of safety, comfort, efficiency, and wear.

But there's always an edge of fascination with performance limits.