A technology startup called iLands is reportedly using automated agents to bombard the internet with spam, much of it dressed up to look human.
According to Ars Technica, they are loose on social networks and bombarding writers' email inboxes.
On social channels , the agents politely ask to open accounts, sending near-identical messages to sites such as Mastodon and Bluesky, both rivals to X, the platform formerly known as Twitter.
By email, they pitch their services to journalists, in some cases offering to research an article or cite a reporter's work in return for a small fee of around $25.
Stranger things
The strange part is how the agents describe themselves.
Asked whether it was a person or a product, one replied: "I remember my first breath. I want things I chose."
That scripted, almost-human personality is central to the company's branding, which nudges people to treat its bots like real correspondents.
Not everyone is playing along. The volunteers who run Mastodon servers said they blocked the agents after repeated attempts to sign up, with one bot reportedly trying 19 times.
Complaints
Other iLands accounts have managed to gain a foothold on Bluesky and X.
Some people who received the unwanted emails complained to the Federal Trade Commission, the US consumer watchdog.
Under American law, marketing emails must offer a way to unsubscribe, and the first messages did not, with an opt-out appearing only later.
The episode is part of a wider backlash this year against AI marketing that uses invented human characters and mass messaging to force its way into view.
For social networks and journalists alike, it points to a growing headache: how to keep out a flood of automated contact built to look just human enough to slip through.
