Credit: Jordan Gloor / How-To Geek

Sign in to your How-To Geek account

Do you keep seeing web pages pop up with a chibi-style drawing of a girl with dog ears and a magnifying glass under the text “Making sure you’re not a bot”? After a few seconds, it probably disappears, and you arrive at the web page you were visiting. What’s going on?

Don’t worry, you don’t have a virus. I was worried about this too, but it turns out that the issue is just that I spend too much time on Linux documentation and other open source projects that are trying to protect their sites from AI-fueled overloading.

You’re running Into Anubis

When you see the dog-eared girl with the magnifying glass, you’re just encountering an Anubis checkpoint. Anubis is a protective layer website owners can apply to their domain that acts as a sort of firewall against web-scraping bots. It’s not malware, and it doesn’t mean there’s anything wrong with your computer.

Anubis was created by developer Xe Iaso in early 2025 in response to the proliferation of web scrapers that consume internet content and feed it into AI model training programs to improve chatbots like ChatGPT and Perplexity AI. These web-scraping bots designed by the likes of Amazon, Meta, and Anthropic are aggressive and deceptive. They ignore traditional requests for bots to stay away and use workarounds to avoid old methods of identifying and blocking bots.

The bonkers amount of traffic they send to public websites can easily burden website hosts, especially small ones, and can slow or shut down websites entirely. That’s why website owners turn to solutions like Anubis to keep those annoying web crawlers at bay.

As reported by 404 Media, it’s popular with open source projects that are mostly volunteer-run and can’t afford to put up with oversize server loads. I spend a lot of time researching Linux and other open source software, so I run into Anubis a lot.

How does Anubis work?

You might think of Anubis as similar to CAPTCHA challenges in that its goal is just to confirm you are human. However, unlike CAPTCHA, there’s no puzzle for you to solve with Anubis. The process takes place entirely behind the scenes, in your browser. Anubis doesn’t check if you’re a human so much as it checks if the browser you’re using is one that humans use.

So Anubis does technically involve puzzles, they’re just puzzles solved by your browser instead of you. Anubis asks Chrome, Firefox, or whatever browser you’re using to run some cryptographic calculations and other tests in the background using JavaScript (at the time of writing). These are calculations that most bots aren’t equipped to run, and if they were, the cost of keeping those bots operational would go through the roof.

This also means that if you disable JavaScript in your browser you probably won’t be able to get past an Anubis checkpoint. JavaScript is an integral part of the user experience on most popular websites these days, which is why it’s enabled by default in most browsers. I’ll note though that the developer has said they are working on ways to make Anubis work even without JavaScript, as some real humans do turn off JavaScript for privacy reasons.

Why not regular CAPTCHA?

You may be wondering at this point why anyone would go through the trouble of developing an entirely new bot firewall if we already have CAPTCHAs. They’re annoying, too, but they work—right?

Not so much anymore. Modern web-scraping bots have CAPTCHA-solving abilities. In fact, it’s a specific issue the developer cited in their blog post explaining Anubis. CAPTCHA can’t offer much protection to website owners anymore. What’s worse, the arms race it creates means you, the human, have to solve increasingly difficult puzzles just to interact with the web.

At least Anubis only makes you sit and wait a few seconds. I prefer that over identifying motorcycles and buses.

Annoyed? Blame our tech overlords

Anubis wouldn’t exist if big tech companies acted with respect. In the days before tools like ChatGPT exploded in popularity, web scrapers and other kinds of bots typically agreed to requests that website owners made to be left alone. That honor system became a thing of the past now that data for AI training has become the subject of a gold rush. Despite site owners’ pleas, major tech companies have been detected hoovering up training data anyway and overloading servers in the process.

So long as tech companies continue to pursue bigger datasets for their AI training, and so long as they do so without respect for website owners, Anubis’ dog-eared girl and other bot firewalls will continue to pop up as you browse the web.