CAPTCHA

You have probably squinted at some wavy, bent letters and numbers on a screen at least once, maybe while buying limited-edition sneakers or booking tickets to a hot concert.

Bots and macros are automated, harmful programs. They can click and type data far faster than any person. If a website had no defenses, a scalper program would grab expensive concert tickets in the blink of an eye, and the message board would fill up with spam ads.

Early CAPTCHAs fixed this by using something only humans were good at: visual perception. Computer programs of that era could not read badly distorted letters. People could pick out letters full of noise and distortion from the background without much trouble.

This gap in visual recognition kept programs out and protected the server's precious resources. Typing the letters every time was a bit of a hassle for users. Still, it worked as a solid shield that filtered out large-scale attacks by malicious macros.

These days, you rarely stare at tangled letters to type them in. It is much more common to tick an "I'm not a robot" checkbox once, or to pick out traffic lights and fire hydrants in street photos.

To be more precise, modern CAPTCHA looks at more than whether you got the quiz right. It also watches behavior signals, like how your mouse cursor moves. A person moving a mouse naturally makes tiny tremors and curves. A simple automated program often zips along a mechanical straight line in an instant. Still, clever bots can imitate human movement, so mouse movement alone is not used to decide.

It is also said that these systems compare your IP address, cookies, how long you stay on a page, and your web browser's settings to give you a risk score. The exact way it decides has not been made public. Everyday actions that mean nothing to a person add up to strong proof of who you are.

Thanks to this, users no longer have to type long strings of characters. A light click is enough to prove to the server that you are a real person.

AI and deep learning have advanced dramatically in recent years. Computer programs can now find crosswalks in photos or read warped letters far faster and more accurately than people.

Once bots' vision began to beat humans, CAPTCHA technology quickly moved away from the old way of handing users a quiz to solve. The best-known example is the "invisible CAPTCHA." It quietly watches how you normally browse a page and turns that into a score for how likely you are to be human.

With this approach, only the very few visitors strongly suspected of being malicious bots are asked for an extra verification step. As a result, most regular users are never interrupted and can use a website's many features smoothly and easily.

CAPTCHA is no longer a gatekeeper who hands you a tedious test paper. It is turning into a smart security system that quietly checks how a guest walks from behind the scenes and then opens the door safely.