“The AI vulnerability detection tool Mythos had a false positive rate of 30% during testing at Palo Alto Networks.”