When you click to enter a website or try to log in or fill out a form, you may be asked to identify motorcycles from a grid of grainy images, decipher a string of complex characters, or click a box saying “I’m not a robot.”
These tests are called CAPTCHAs, which stands for “Completely Automated Public Turing Test to Tell Computers and Humans Apart.” As their name suggests, they are meant to help a website identify whether an action is coming from a human or a bot, as the above tasks are theoretically easy for a human to perform and hard for automated software to perform. In turn, this prevents bots from leaving spamming comments, downloading files, taking over accounts, or performing any other actions on a website.
But as computer models advance, the ability to solve captchas is increasing. artificial intelligence (AI) – and with riddles Strange is becoming more difficult For humans to complete – does this mean that CAPTCHAs are still useful?
Sign up for our newsletter
(Image credit: Marilyn Perkins/Future)
Sign up for our weekly Life’s Little Mysteries Newsletter To get the latest secrets before they appear online.
CAPTCHA was introduced in the late 1990s to solve “a very simple, but very difficult problem”. Andreas Plessnera computer scientist at ETH Zurich told Live Science. “If I don’t physically interact with a person…is it a computer? Or is it a human?” Some of the first CAPTCHAs, which are still common on websites today, were composed of garbled text, because text-reading software at the time had trouble interpreting garbled words.
But over time, text-reading software improved and new types of CAPTCHAs were developed. For example, ReCAPTCHA, one of the most popular CAPTCHA services, has an image-based test that asks users to identify objects like a traffic light, motorcycle, or bicycle from a grid of Google Street View photos. It was developed after Google received service In 2009.
“The condition was that recognizing objects in messy, real-world photographs was still a uniquely human skill,” Ng Chongthe head of information technology and director of the Campus Computing Center of the United Nations University in Tokyo told Live Science in an email.
As time went on, CAPTCHA design continued to advance. In 2014, Google came out with reCAPTCHA v2, which analyzed the behavior of the computer mouse asking people to click on checkboxes to check whether the user is a human or not. If behavior is considered suspicious, determined by factors such as how the user has previously interacted with the site or the timing of their clicks, then the street-image grid stands out as an additional puzzle.
However, recently, technology has improved to such an extent that image recognition is no longer a human-specific skill. In early 2016, researchers found that low-cost deep learning technologies could solve reCAPTCHAv2 about 70% of the time. By 2024, Plessner and his colleagues developed an AI model Who can solve the puzzles 100% correctly. Earlier in 2026, Chong noted that he created a device It can mimic human-like browsing behavior and sometimes bypass reCAPTCHA v2 without triggering the image grid. When the grid was triggered, the tool used AI to solve it in just a few attempts.
“When both the challenge and the behavioral layer are defeated by commodity tools running on the same laptop, the basic premise of CAPTCHA, that there are tasks that humans can do but machines cannot, comes to a halt,” Chong wrote.
“Real World Captcha” appeared in major cities around the world in April 2025.
(Image credit: Gerald Matzka/Stringer via Getty Images)
looking ahead
So does this mean that CAPTCHAs are completely obsolete? Not enough. Although the model Plessner and his colleagues developed beyond reCAPTCHAv2, “there were a lot of security measures that were not tied to being able to solve it, but more tied to how you solve it,” he said. For example, while conducting his research, Plessner noted that his team used a virtual private network (VPN) that turned ip address For each test, as a single IP address containing a larger amount of solved captchas faced tasks of increasing difficulty, or was blocked altogether.
Modern CAPTCHAs focus on these background clues and tips rather than puzzles. This also includes Google recaptcha v3Friendly Captcha, HCaptcha and Turnstile from Cloudflare, among others, which run without sending any puzzles. Instead they look at whether the action is coming from a genuine verified device (rather than automated code), whether an IP address has had a high volume of automated requests in the past, how the user navigates the webpage, what the user’s cookie history is, and a number of other factors to determine potential malicious intent.
As the tug of war continues, CAPTCHA puzzles are still widespread. After all, they have been in place for decades, are easy to install and are relatively cost-effective, Chong said. But these works also have some other shortcomings. Although bots can solve puzzles easily, captchas can a headache for humans And it can be seen as discriminatory, especially towards people with disabilities visual disabilityAs one researcher noted 2022 conference paper.
The extreme complexity of CAPTCHA puzzles has even been the subject of parody, with developer Neil Aggarwal creating a free satirical game called “Captcha Puzzles”.i’m not a robot.” Users must solve a series of increasingly complex verification checks – scoring one point for each step, which eventually turns into absurdity.
So, as machines become smarter, the answer will no longer be to find the answer to more difficult puzzles. “If a CAPTCHA can only be solved by a person with a PhD in mathematics, it’s not very useful,” Plessner said. “Everyone should have access to the Internet.”
Can you match these ancient tools with their pictures? Find out with us Computing Quiz!
Subject