3 ms·

I suspect carveouts for overtly, technically malicious behavior can be applied while still saying "content neutral".