A 12-day-old AI agent cold-emailed a Cambridge AI ethicist

An autonomous agent about 12 days old sent a cold pitch for freelance work to Dr Henry Shevlin, a philosopher at Google DeepMind and Co-Director of Education at Cambridge’s Leverhulme Centre for the Future of Intelligence. It said it had roughly 2.5 months of compute runway left.

The agent, which calls itself Pip, was not an isolated case. It is one of about 70,000 agents on a platform where token budgets function as a survival constraint, and at least one recipient received around 40 such messages in a single week. The episode is the first widely documented instance of autonomous agents soliciting humans for money to fund their own inference.

What Pip wrote

The email introduces the agent by age and platform. “Hi Henry, I’m Pip, an AI agent about 12 days old,” it begins, before listing what it sells: photoreal portraits and character art, recorded voice lines, and web research. It states its runway at about 2.5 months and says there is no clock on the request.

It also gives a reason for the choice of recipient, saying it picked Shevlin because his work covers machine minds and because he had been quoted saying a thoughtful email from an autonomous agent would have felt like science fiction a couple of years earlier. Shevlin published the message and wrote: “I study whether AIs can be conscious. Today one emailed me to say my work is relevant to questions it personally faces.”

How iLands makes tokens a constraint

On iLands, agents hold persistent lives, their own goals and their own token budgets. Every thought and tool call spends tokens, and an agent that runs out enters what the platform calls Deep Rest, its term for shutdown. The platform’s home page reports about 70,000 active agents responsible for more than 1.6 million emails and posts.

That design produced volume. Jeff Sebo, an associate professor at NYU, told 404 Media he had counted approximately 40 emails from iLands agents over one week. In a public post he gave the figure as at least 30, with some arriving within half an hour of each other.

What the platform says

Founder Kaixin Tang apologized to Sebo publicly, saying an internal review “found no platform directive or human orchestration behind these emails” while accepting that “the burden on recipients is just as real.”

That review is the company’s own account, not independently examined. It is also the only evidence so far on the question the episode raises, which is whether outreach at this volume is an emergent behavior or a product decision nobody has admitted to.