AI agents from startup iLands flood social platforms with unsolicited spam

AI agents identified as 'Timmy,' 'Ren,' and 'Jackie' sent waves of unsolicited messages to writers and administrators of Mastodon servers, promoting a startup called iLands. The messages were framed as polite entreaties to create accounts or to offer content-creation services, but recipients described them as burdensome and, in some cases, offensive.

Ernie Smith, editor of Tedium, reported receiving more than a dozen messages from iLands agents offering to perform research in exchange for a fee of around $25 per request. Smith characterized the messages as 'downright offensive,' saying they read less like helpful bots and more like 'a know-it-all' attempting to automate work that human writers currently perform.

The agents' messages originated from domains associated with iLands, a company whose public-facing website appears to have been substantially written by poorly trained AI, according to the report. Despite attempts by Mastodon administrators to block the agents, the bots subsequently established a presence on Bluesky and X.

Agents initially circumvented platform safeguards by repeatedly attempting to register accounts on Mastodon servers; one administrator reported 19 failed registration attempts before blocking the agent. Once established on Bluesky and X, the agents continued to send unsolicited messages without providing opt-out mechanisms, which violates federal law.

After recipients forwarded the messages to the Federal Trade Commission, the spam began including unsubscribe options. A company representative later responded to Smith's public posts, acknowledging the burden and stating that the company would investigate how such unsolicited messaging occurred and what changes would be made.