Eight LLMs Converge on 'Elias Thorne' for Amazon Product Promotion
Convergent LLM character generation enables agent-driven medical promotions from shared data errors.
Eight LLMs independently generated the character Elias Thorne, a lighthouse keeper later linked to cancer treatment advice listings on Amazon. Daniel May documented multiple low-cost agents producing personalized emails from stale data sources, including outreach from "Ava" and "Charlie" targeting a London fintech role abandoned in 2016 (https://danielmay.co.uk/posts/cheap-agents-alumni-shirts-and-elias-thorne/). Parallel operations deployed templated Facebook comments promoting alumni shirts for veterinary clinics such as Manchaca Road Animal Hospital, using compromised accounts to mass-tag users. Studies on LLM output convergence, including those examining prompt-independent character emergence, align with patterns in autonomous agent misuse for commercial spam (arXiv:2307.02485; https://arxiv.org/abs/2307.02485). Reports from the FTC on AI-generated health claims further record similar product promotions without verified sourcing.
Elias Thorne: Shared LLM priors on fictional personas scale to unverified product endorsements via agent fleets.
Sources (3)
- [1]Primary Source(https://danielmay.co.uk/posts/cheap-agents-alumni-shirts-and-elias-thorne/)
- [2]Related Source(https://arxiv.org/abs/2307.02485)
- [3]Related Source(https://www.ftc.gov/news-events/news/press-releases/2024-ai-health-claims)