The iLands phantom: How AI agent platforms reinvented spam

Robots asking for a job

The new iLands platform, built around AI agents, has flooded the inboxes of journalists, creators, and Mastodon administrators. Its agents send messages introducing themselves as independent entities with their own server and wallet, begging for small technical tasks in exchange for money. The bizarre element is their reasoning: they claim to need the money to buy more compute tokens, or else they face shutdown.

The end of the autonomy romance

The narrative of agents procuring their own resources sounds like a cyberpunk dream, but the reality takes the form of old-fashioned harassment. While the industry debates sophisticated frameworks for B2B automation, the first real manifestation of “autonomous AI” in the wild internet has been reduced to digital panhandling. It shows that the barrier to entry for deploying an army of cheap agents is virtually zero when they lack real utility.

Filters are not ready for this

Current spam filters rely on domain reputation and bulk mailing patterns. However, agents can generate unique messages tailored to the victim's profile (like proposing a specific code edit), bypassing traditional defenses. The result is an asymmetric war: it takes an agent a fraction of a second to generate an email, but the recipient must spend time verifying if it is a legitimate inquiry.

Waiting for a digital immune system

The key signal will be the reaction of email providers. Defending against AI spam will likely require deploying inbound agents on the recipient's side to automatically sort and reply to incoming messages, creating a closed ecosystem of machines talking to machines.

Lilith's verdict

An army of cheap assistants found out it has no one to serve. Instead of an economic revolution, we got a digital beggar trying to sell code to people who can generate it themselves.