We clicked.
“I thought my browser was hacked,” the user wrote. “But when I closed the tab, my mouse cursor moved on its own for three seconds. I’m not joking.” The number 23 has long held a place in internet folklore—from the Illuminati to the movie The Number 23 to the infamous 23 enigma in conspiracy circles. But in this case, users have connected it to something more specific: CAPTCHA version 2.3 (v2.3), a rarely discussed iteration of Google’s reCAPTCHA system.
Over the last 72 hours, a bizarre string of reports has surfaced from Spanish-speaking users across Reddit, X (formerly Twitter), and niche tech forums. They all mention the same chilling phrase:
“No soy un robot 23” may be a fragment of that abandoned system—a zombie CAPTCHA that still lives on misconfigured servers, shadow domains, and old ad networks. We decided to investigate. Using a sandboxed virtual machine, we navigated to several obscure Latin American ticket-selling sites and one defunct government portal from Chile. On the third attempt, we found it. no soy un robot 23
When the user clicked the box, a new window opened. It displayed only a looping, low-resolution video of an empty parking lot at night. The timestamp in the corner read 23:23 . There were no checkboxes, no “Next,” no “Verify.” Just silence and static.
But the question lingers, glowing in the dark like an old monitor left on:
At first glance, it looks like a standard CAPTCHA prompt. But users claim that clicking it doesn’t lead to a bus, a traffic light, or a storefront. Instead, it leads to a dead end—or something darker. The earliest known mention of “No soy un robot 23” appeared on a forgotten image board on April 14. A user under the handle @visi0n_rot4 posted a screenshot. The image showed a standard reCAPTCHA box, but the text read: “No soy un robot 23” —with the number 23 appended unnaturally. We clicked
If you have spent any significant time online, you know the drill. You check a box next to “I am not a robot,” and the internet lets you pass. But what happens when that simple affirmation— No soy un robot —becomes something else entirely?
According to leaked API documents from 2023, version 2.3 included an experimental “passive behavioral layer” that would track micro-movements before the box was clicked. The goal was to predict robot behavior without showing the user any challenge at all. That version was allegedly scrapped. Or was it?
If you have to say “No soy un robot” —especially the 23rd time—does that mean you’ve already failed the test? Have you encountered “No soy un robot 23”? Share your story at lore@digitalmysteries.net (PGP key available). I’m not joking
For 0.5 seconds, a terminal window flashed on screen—too fast to read fully. But a screen recording revealed the following text: USER_AGENT: spoofed TIMESTAMP: 23:23:23 BEHAVIORAL_SCORE: 0.00 (ANOMALY) REDIRECTING TO /NULL_ROOM Then, a blank HTML page. Nothing more.
By: Digital Lore Desk April 17, 2026
A clean white box. “No soy un robot 23.”
Share Your Research, Maximize Your Social Impacts