News for hackers. Stuff that matters.
Re:Sob story about dead grandma tricks Microsoft AI into sol... by msm_ on Monday October 02, 2023 @05:45PM (#37744898)
Most CAPTCHAs are already solvable automatically. Usually there's a rate limitter as a second line of defense, and also some heuristics that detect bot-like behaviour (user keeps upvoting posts of certain users without even reading them and uses API in a otherwise non-standard way? Hmmm, throw more CAPTCHas at them and ultimately ban them). Finally, recaptcha and (probably cloudflare's captcha?) tracks wayy more than just how correct you are in recognising street signs, and correlates this to your overall network activity.

You can't rely on just CAPTCHAs anyway, because mechanical Turks are too cheap compared to the damage they can do.

Re:Sob story about dead grandma tricks Microsoft AI into sol... by bentcorner on Monday October 02, 2023 @06:29PM (#37745316)
Maybe we end up taking the problem to a deeper level - for some accounts the true test if they are human is if they fail a captcha test.
Re:Sob story about dead grandma tricks Microsoft AI into sol... by follower on Monday October 02, 2023 @07:25PM (#37745840)
...or if they refuse to dance like a monkey and close the window instead.