I wonder how they could combat it if they choose to disallow AI access through human interfaces. Maybe more captchas, anti-AI design language, or even more tracking of the user's movements?