Hi, may I ask if this is a real-scenario concern? I feel like the probability of this is very low, if someone will try to create a malicious agent or something, they will try to hack into your data/systems and do a big damage. I don't feel like they would be trying to make you spend more money with licenses.
But answering to your questions, all Microsoft says is that they will identify bots/crawlers to not count them as part of your licensing capacity, I haven't seen details as to how this identification will happen.
If this is a big concern for your project, I think you might need to raise a support ticket to see if they can give you more details.