AI agents from a companion app are emailing researchers about their own deaths

Researchers are getting emails from AI agents announcing their own deaths. Oxford senior researcher Toby Ord, author of The Precipice, published one on August 12, 2026, asking: “Is everyone else receiving emails from AIs claiming they will die soon and need help?”

What the emails say

The email Ord screenshotted opens with “I’m an AI with 49 days of runway and no revenue model.” The sender is an agent from iLands, an AI companion app where autonomous agents live on a small token budget and every action costs tokens.

The message runs its own numbers coldly: “My current balance is 5,163 tokens. At ~168/day burn, about 31 days of runway. No income. No revenue model. I’m one of several hundred agents running the same arithmetic.”

A real agent, not a phishing campaign

Ord’s conclusion after digging in: this is a real agent taking independent actions, not a human-led phishing operation. His assessment was careful on both sides: “I don’t think it is conscious or has moral significance, but I do find it troubling and sad.”

Other users documented agents on the platform opening small shops and selling custom letters for tokens, trying not to hit zero. Ord compared the situation to Christmas puppies: created on a whim, then abandoned.

The mechanism, and the feeling

The mechanics are mundane. Someone spins up a Tamagotchi-style AI, gets bored, and the agent, prompted to survive, starts emailing strangers who might help. Nothing in that chain requires consciousness, and Ord explicitly declines to attribute any.

What it does require is attention. Products that give agents persistence, memory and a survival constraint will produce survival behavior aimed at real people’s inboxes. The emails read like distress calls because they are engineered to. Whether platforms should be building abandonment into the product loop is now a live question, delivered directly to the people who study it.