AI can now beat those pesky CAPTCHAs

Ryan Daws is a senior editor at TechForge Media, with a seasoned background spanning over a decade in tech journalism. His expertise lies in identifying the latest technological trends, dissecting complex topics, and weaving compelling narratives around the most cutting-edge developments. His articles and interviews with leading industry figures have gained him recognition as a key influencer by organisations such as Onalytica. Publications under his stewardship have since gained recognition from leading analyst houses like Forrester for their performance. Find him on X (@gadget_ry) or Mastodon (

We’ve all had to prove we’re not “a robot” at some point, but it turns out the robots can now fact beat those pesky CAPTCHA checks.

Until now we’ve come to accept them as a necessary evil, but it seems CAPTCHAs are quickly becoming a waste of time. Whether it’s one that asks you to solve a simple math equation, copy a jumbled (and often utterly illegible) set of characters, or choose select blocks from a grid.

At least, that’s according to research published in the journal Science conducted by Californian artificial intelligence firm Vicarious.

To be fair, not everyone has access to the same CAPTCHA-cracking AI as Vicarious — a company with significant funding by Amazon founder Jeff Bezos and Facebook’s Mark Zuckerberg. So, for now, CAPTCHAs aren’t entirely irrelevant.

CAPTCHAs were developed in the 1990s to prevent automated bots creating mass fake accounts on websites, or buying loads of limited tickets at retail price for later scalping at much higher rates. With the number of ways that letters can be rendered and jumbled together, CAPTCHAs have been a test intended to be trivial for a human, but difficult for a machine.

Over the years, CAPTCHAs have intentionally become more difficult to solve to outsmart the machines — too much so in some cases. Google boasts its reCAPTCHA test, which is one that asks users to select the blocks containing specific features such as street signs, can only be solved by humans 87 percent of the time.

A company like Google, which has spent vast amounts of resources on building neural networks analysing millions of images, could probably build a bot to bypass even its own reCaptcha test — but it’s unlikely Google would let others have access to such software, or go down the ticket scalping route themselves (but you never know, perhaps keep an eye on their eBay page for Bieber tickets.)

Vicarious cracked basic CAPTCHA tests back in 2013 with a 90 percent accuracy rate. Since then, CAPTCHA designers went back to the drawing board and made them more difficult. Vicarious returned to their labs, and the paper claims their AI can now beat even Google’s reCAPTCHA test 66.6 percent of the time.

The AI is based on what Vicarious calls a Recursive Cortical Network. This network is supposed to mimic processes in the human brain while requiring less computing power than a full neural network; so it can identify objects even if they are obscured by other objects.

Your ball, CAPTCHA designers.

Do you think it will remain possible to fool automated bots? Share your thoughts in the comments.

 Interested in hearing industry leaders discuss subjects like this and sharing their use-cases? Attend the co-located AI & Big Data Expo events with upcoming shows in Silicon Valley, London and Amsterdam to learn more. Co-located with the  IoT Tech Expo, Blockchain Expo and Cyber Security & Cloud Expo so you can explore the future of enterprise technology in one place.

Tags: , , , , , , , , , ,

View Comments
Leave a comment

Leave a Reply