It sounds like you're implying that it is impossible to create a system that makes penalties scale instead of just being a fee because the system will always be gamed.
Do you believe that or did I misunderstand you?
Whereas the proposal is typically to make the penalties scale with entity size, which a) is actually easier to game than the scope of the harmful activity, and b) does a bunch of things that make no sense, like increasing the amount of the fine for an oil spill if the company also generates additional revenue from producing solar panels.
The reason we run into trouble here usually isn't that the penalties are wrong, it's that the enforcement is weak. If you can steal $10 billion with only a 10% chance of enforcement that requires you to give back $30 billion then your expected value is an $8 billion gain. But that has nothing to do with the scale of the business. The exact same thing happens to with a company that steals $10,000 with only a 10% chance of enforcement.
And using disproportionate penalties doesn't really fix it either, because then they just roll the dice and file for bankruptcy if they encounter enforcement. You can't fine them more money than they actually have. When the rate of enforcement is too low, that isn't enough to make them comply, because a small chance of -100% is profitable against a large chance of +50%. Whereas if enforcement is consistent then disproportionate penalties aren't even necessary.
Being able to game a naive implementation of such scaling is not an argument for not scaling penalties but for designing the concrete implementation to be hard to game and/or applying even higher penalties when someone has been found to try to game the system - which is something that should be done more in general.
We can absolutely make it so that people hiring private chauffeurs share in the liability for breaking traffic laws just like we can absolutely hold CEOs liable for illegal activities done by employees under their purview.