The Ultimate Guide to Spammy Ultimate Guides
This swarm of AI-generated blogs is an exercise in quantity, not quality
Over the last few years, the widespread availability of increasingly sophisticated large language models has served as a boon to spammers who seek to flood the internet with high volumes of low-quality articles. In one recent instance, a network of dozens of topic-specific websites has begun publishing large numbers of highly repetitive articles on a daily basis, each written by an alleged author with an oddly generic biography. The websites in question have multiple similarities suggesting that they are operated by the same person or entity, and all of the sites are linked with varying degrees of frequency by a recently created Bluesky account with a default avatar, bloglink.bsky.social (“Fern Hart”).
This network of spam blogs consists of at least 44 distinct websites, with a wide range of themes including beekeeping, cat care, guitar technique, air fryers, and knife sharpening. The domains were registered in two waves, with the first 26 domains registered between October 16th and November 12th, 2025, and the remaining 18 between May 28th and June 11th, 2026. At the present time, all 44 domains resolve to Cloudflare IP addresses.
Each of the spam blogs publishes quite prolifically, with numerous similarly-titled articles such as “Ultimate Cat Body Language Guide (2025)”, “Ultimate Guide to Reading Cat Body Language (2025)” and “Ultimate Guide to Cat Body Language (2025)” posted per day. Confusingly, many of the titles include the year 2025 despite the associated articles being generated in 2026. The high volume of material, along with the self-similar nature of the content, suggests that the articles are being generated via AI language models using highly repetitive prompts.
In another potential sign of artificial generation, the content of the articles is extremely similar from post to post and from site to site, with certain elements showing up repeatedly regardless of the purported topic. For example, the first paragraph of nearly every article on each of the websites contains a reference to the number of years of experience the alleged author allegedly has in some field of study or work. (“In my twelve years of teaching”, “after 12 years of working with over 800 cats”, etc.) The overall writing style is extremely consistent across all 44 sites.
The alleged authors of each spammy blog have extremely similar biographies with extremely similar formatting. These biographies have links to LinkedIn and Twitter (now known as X) at the bottom, but in a hilarious bit of ineptitude, these links do not work, and many of the supposed Twitter handles exceed the platform’s 15 character username length limit. (Example: @amaraokonkwo_cats.)
The blog posts produced by this spam operation use a mix of AI-generated images and stock photos as preview/thumbnail images. In some cases, however, these images do not match the content of the articles to which they are attached. For example, the thumbnail for a June 10th article about cat body language includes numerous depictions of human body language, and the image for a June 11th metronome guide shows a green apple covered in water droplets. Given the speed at which the articles are published, these mismatches are likely the result of an automated process for selecting header images going off the rails.
Additional evidence of automation surfaces when one examines the posting schedule of bloglink.bsky.social, the Bluesky account that has thus far shared over one thousand links to the various spam blogs in the network over the course of just ten days. This account generally operates around the clock with insufficient gaps for most human operators to sleep, and for the first eight days, its posting activity spiked at precisely the same times each day. The pattern has changed somewhat as of June 12th, with a significant increase in posting volume and more variation in schedule; time will tell whether this change in posting behavior is permanent.
While I was taking screenshots to use in this article, I observed the bloglink.bsky.social account change its display name in real time on June 11th, 2026. The screenshots in the collage above were taken three minutes apart; in the first, the account has the same display name as handle, “bloglink.bsky.social”. In the second screenshot, the display name has been changed to “Fern Hart”, presumably to create the impression that the account is run by a person with that name, rather than merely being an exercise in spam.
Despite posting a large number of links to spammy AI-generated blogs, the bloglink.bsky.social account thus far has not managed to build much of an audience, with only 15 followers as of June 12th, 2026. Even this number is inflated, as the account’s most recent three followers are strangely similar porn spam accounts named “Emma Wilson”.
