You have to be very careful when writing these laws. You can either write them from the perspective of saying platforms are responsible for disclosing risk and demonstrating progress to reduce risk, or you can write them to say that every instance of hate speech has a penalty. The challenge with the latter kind, where you say you have a zero-tolerance perspective—no hate speech—is that computers can't accurately and reliably identify what is okay and what is not okay. What it will mean is things like erasing trans people from the Internet, because it can't tell whether a comment is hateful. It would mean erasing religious minorities: Can you talk about religion confidently if there is a $40,000 violation for the platform?
As long as the law stays within the bounds of saying you must disclose what risks you believe exist and demonstrate your progress, that can be okay, but we need to be careful not to believe that we can erase hate from these platforms without accepting that we will also erase lots of legitimate speech, because the computers are just not that smart.