Terrible Website of the Day

Some AI company I've never heard of made a way for webmasters to opt out of Web crawlers used to train AI models. The idea is fine and the implementation is exactly the same as robots.txt, which is also fine. What is not fine, however, is the Web page they created to promote it.

The first offense is the message when you first visit the page, "You need to enable JavaScript to run this app." Usually, when I see that message, I just close the tab. This time, I decided to push forward and enable JavaScript.

The second offense is the fact that the page is indeed a page comprised mostly of text, not an "app" as previously described. It also does not contain an example of an ai.txt file anywhere, which is the only thing I actually wanted to see. Instead, there are a few bullet points that were probably written by a language model themselves and an embedded YouTube video with an incredibly slow talking voice that also doesn't show an example of an ai.txt file.

There is also a generator to create your own ai.txt file. Apparently, if you want to use the generator, which is entirely client-side JavaScript, you must agree to the terms of service. The link to the TOS is not real. It doesn't show me the URL when I hover over it so it's using an event listener to open a new tab when it's clicked.

When I click on the line of text resembling a hyperlink, it opens another page comprised entirely of text that will not display a single thing without enabling first-party JavaScript and XHR from a third-party website. What a clown show!

Appendix: More Offenses

You need to scroll down a full page to get any actual information. It's just a giant, irrelevant picture with a marketing tagline and a couple of buttons to scroll down. This is an inexcusable pattern in Web design that should be called out whenever it appears. Credit @lyse@lyse.isobeef.org on twtxt. I really should have picked up on this.

There are a couple things I've realized since writing this post.

1. The page has no DOCTYPE, so it's rendered in "Quirks mode" in Web browsers. Quirks Mode wasn't acceptable on Geocities, let alone a website in 2024 made by a real company with paid staff.
2. The page does not tell you what their User-Agent string is, which is pertinent information. It looks like it's Spawning-AI according to @lyse@lyse.isobeef.org's logs.

Updated 2024-01-25: Added appendix, minor wording adjustments.

[ ← Systemd: Unmount NFS before killing the Wireguard interface | Notes Index | On Website Builders → ]