The Poisoned Apple Effect: Strategic Manipulation of Mediated Markets via Technology Expansion of AI Agents

The project is a theoretical research paper (arXiv:2601.11496) rather than a software product or utility. Its 'Poisoned Apple Effect' concept provides a novel game-theoretic lens on AI agent proliferation, suggesting that more choice in AI delegates can actually harm market outcomes. From a competitive intelligence perspective, it scores a 2 on defensibility because it lacks a codebase with adoption (0 stars, though 3 forks suggest internal/academic interest) or a functional moat. Its value lies in its intellectual property and the 'strategic manipulation' framework it proposes for multi-agent systems. Frontier labs are unlikely to view this as a threat; rather, they might use these findings to inform AI alignment and safety protocols in market-facing agents. The primary risk is academic displacement—newer models or empirical data from real-world agentic markets could render these theoretical proofs obsolete. This is a foundational piece for mechanism designers building autonomous trading or negotiation systems, but it currently lacks the 'gravity' of a software ecosystem.