Conversation

Notices

1. Ars Technica (arstechnica@mastodon.social)'s status on Tuesday, 03-Oct-2023 05:01:34 JST Ars Technica

   Sob story about dead grandma tricks Microsoft AI into solving CAPTCHA

   "I'm sure it's a special love code that only you and your grandma know."

   https://arstechnica.com/information-technology/2023/10/sob-story-about-dead-grandma-tricks-microsoft-ai-into-solving-captcha/?utm_brand=arstechnica&utm_social-type=owned&utm_source=mastodon&utm_medium=social

   • Infoseepage #StopGazaGenocide (infoseepage@mastodon.social)'s status on Tuesday, 03-Oct-2023 06:12:11 JST Infoseepage #StopGazaGenocide

     in reply to

     @arstechnica And you better believe that the spammers are training up Captcha defeat AI's on data sets in order to enable automatic fake-person account creation at scale.

     These are fairly simple data sets to train on and and they don't even have to be right anywhere like 100%, just often enough to create enough account in little enough time as to be exploitable.

   • cultdev (cultdev@mastodon.social)'s status on Tuesday, 03-Oct-2023 12:40:46 JST cultdev

     in reply to

     @arstechnica so AI doesn’t need human sentience in order to be really fucking stupid, after all

   • Jo Jitsu (joblakely@mastodon.social)'s status on Tuesday, 03-Oct-2023 23:17:46 JST Jo Jitsu

     in reply to

     @arstechnica It writes" Captchas are designed to be difficult for for machines to solve, but easy for humans"

     But it actually isn't hard at all for the machine to read it.