Captcha is a security feature that ensures that anyone who wants to access certain parts of a website is a real human and not a robot that wants to steal content from the site.
The Captcha system usually comes with a question that needs to be answered by typing the pictures requested by the question, and if successful, you can go to the next page, and if not, you will have to answer it until successful.
Recently, it has been reported that existing artificial intelligence technology seems to have succeeded in overcoming this Captcha test, specifically the one using reCaptchav2 technology by Google. Through a newly released paper, some researchers seem to have taught an artificial intelligence system how to see images accurately using 14 thousand pictures.
Once trained, it was found that this artificial intelligence system could pass the ReCaptcha v2 test with 100 percent accuracy, something never seen before.
Even so, this is only seen to be true with the ReCaptcha v2 system, where the ReCaptcha v3 test, which requires more dynamic user input (such as pulling an image to the right location) has yet to be overcome by machine learning.
This research attempts to show that although artificial intelligence technology can be trained to overcome this old Captcha test, it is still a good security system to prevent access to bots on the internet that are developed to steal content.