One frustrating website security feature just became useless

The web's best loved test of whether you're a human being just became redundant - according to US technology firm Vicarious.

The San Francisco based company says that its software can break tests which have users input a distorted string of words and symbols - known as CAPTCHAs - and they say they can do it with high rates of accuracy.

Vicarious says that CAPTCHAs can be considered broken if their precision exceeds one per cent - and they say "the Vicarious AI achieves success rates up to 90% on modern CAPTCHAs from Google, Yahoo, PayPal,, and others."

But some aren't convinced that the Vicarious AI will change that much.

Luis von Ahn of Carnegie Mellon University, one of the developers behind CAPTCHAs, told Reuters that even if letter based tests have been cracked that "CAPTCHAs will use something else, like pictures".

And according to Greg Mori of Simon Fraser University - a more human crack already exists. He says that many CAPTCHAs "are broken by paying people" to do the job. Mori claims that "for 50 cents an hour, you can get someone to break seven per minute".