I don't remember the name, but I've seen a Captcha that asks you stuff like "click on the Zerki", "select the Sploinq" with AI images and I have no idea what shape most looks like a Zerki (there are more than one shape with hard edges!) or a Sploinq. And I have a linguistics minor. :(

hCaptcha has absolutely wanted people to click on objects that do not exist even on a conceptual level.

https://www.vice.com/en/article/xgwy5n/captcha-is-asking-use...

hCaptcha is garbage. Also, they use some blockchain/token stuff for no reason, at least the last time I looked.

edit: ah they seemed to have removed all traces of the token stuff from their website, except some small remains in the docs - https://docs.hcaptcha.com/faq/#what-is-one-hmt-worth . I guess they pivoted away, reasonably.

Still better than the google malware that lets you through based on how well they can track you around the web rather than based on a legitimately solvable challenge. At least, I noticed that I never saw anyone but myself getting any challenge at all, let alone the worst CAPTCHAs that google has to offer. Same IP address, the only difference that I could find is not having "auto delete cookies" installed in the browser (wipes localstorage etc. after you closed all tabs of non-whitelisted sites for some seconds).

hCAPTCHA has always been easily solvable for me, haven't seen the one you linked yet but that also looks quite trivial

Ideally one just doesn't use CAPTCHAs at all, but my colleagues disagree and so that's unfortunately the company policy to recommend against login brute force and such