Anubis works because AI crawlers do very little requests from an ip address to bypass rate-limiting. Last year they could still be blocked by ip range, but now

Anubis works because AI crawlers do very little requests from an ip address to bypass rate-limiting. Last year they could still be blocked by ip range, but now the requests are from so many different networks that doesn't work anymore. Doing the proof-of-work for every request is apparently too much work for them. Crawlers using a single ip, or multiple ips from a single range are easily identifiable and rate-limited.