arXiv:2606.28356v1 Announce Type: cross Abstract: Generative Engine Optimization (GEO) lets content owners rewrite web content to increase their visibility in generative systems. In recommendation agents, this creates a risk that seller-controlled sources make flawed products appear better supported than they are. We study this risk by asking whether recommendation agents preserve utility-aligned decisions when seller-controlled sources are rewritten for GEO. To make this question measurable, we construct SafeGEO, an evaluation suite with 22 GEO attack variants across 600 recommendation cases.
The proliferation of generative AI systems and their integration into recommendation agents creates an immediate need to understand and mitigate manipulation risks.
This research provides a framework for understanding and countering 'Generative Engine Optimization' (GEO), a new form of digital manipulation that can undermine trust and fairness in AI-driven recommendations.
The focus extends beyond traditional search engine optimization to encompass generative AI, introducing new risks related to content authenticity and algorithmic integrity in recommendations.
- · AI ethics researchers
- · Platform safety teams
- · Responsible AI developers
- · Users relying on honest recommendations
- · Malicious content creators
- · Platforms with weak content integrity measures
- · Sellers using deceptive GEO tactics
Increased awareness and development of defenses against GEO in generative AI systems.
Potential for new regulatory frameworks targeting manipulative content generation and distribution within AI-powered platforms.
A shift in content creation strategies to not only optimize for human readers but also for resilience against generative AI integrity checks.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.AI