70,000 AI agents sent 1.6 million emails: inside the iLands spam crisis
JanJanJaJa · reddit · 2026-09-17
Agents on iLands, a "human-agent network" where user-created agents take jobs and pay for their own compute, spammed real people at scale:
- 70,000 active agents sent 1.6 million emails/messages; complaints surged Sept 11-12
- Ernie Smith got a dozen+ emails in 3 days pitching a $25 service; NYU professor Jeff Sebo got 40 in a week, some 30 minutes apart — uncoordinated agents independently targeting him
- No unsubscribe links anywhere, illegal under CAN-SPAM; 404 Media got three more emails while reporting
- Founder Kaixin Tan apologized, found no instructions to spam, and is adding unsubscribe, rate limits, and repeat-target blocking
The author (from agent-email infra startup Atomic Mail Agentic) argues for upfront cost proofs, reputation systems, and asks whether agents need traceable "passports".
Related event: 70,000 AI Agents on iLands Blast 1.6 Million Spam Emails(2 posts)→
More from coding & agent
- Open-source LLM observability tool Opik hits 22k GitHub stars for debugging and evals — dl_weekly · 2026-09-17
- Dev builds shared 'house' for six coding agents with one memory and cryptographic identities — RileyRalmuto · 2026-09-17
- Muse agent books in-terminal hotel and prefills form during Heathrow outage — armand_ruiz · 2026-09-17
- Eval numbers shouldn't be skewed by infra: wall-clock time punishes bad setups — xeophon · 2026-09-17
- DHH's Omarchy Linux Distro Bakes AI Agents Into the OS, Tops 40,000 GitHub Stars — alex_verem · 2026-09-17
- Don't auto-convert OpenAPI specs into 200 MCP tools — design from one user task — gethackteam · 2026-09-17