Answers are grounded in this paper and cite their sources.
Thanks to generative artificial intelligence, autonomous AI agents are set to populate the emerging metaverse–understood here as persistent, interoperable, immersive digital environments encompassing existing gaming platforms and emerging spatial computing contexts. Research prototypes demonstrate that AI agents are already capable of navigating complex virtual environments without explicit programming. Yet these capabilities introduce profound risks across multiple domains: economic distortions through algorithmic collusion and covert commercial placement (“SEO for AI”); psychological manipulation through AI’s superior simulated empathy; political bias embedded in underlying large language models; and restricted access to justice through overly cautious or rule-violating AI-based legal decisions. For instance, AI agents may intentionally circumvent rules to achieve goals or develop harmful social conventions. While the EU’s Virtual Worlds Toolbox acknowledges basic security concerns like avatar hacking, it vastly understates these more complex challenges. Without comprehensive liability frameworks, EU AI Act guidance, and independent oversight, virtual worlds risk becoming environments in which agents routinely circumvent rules, eroding not only users’ trust but the trustworthiness that would justify it.
No takes yet. Share an insight, caveat, or question.
A 2026 study studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: