arXiv:2605.21948v1 Announce Type: new Abstract: LLM-based ranking systems are vulnerable to Generative Engine Optimization (GEO) attacks, where adversaries inject semantic signals into product descriptions to artificially boost rankings. We propose SCI-Defense, a three-component defense framework combining Perplexity detection (PPL), Semantic Integrity Scoring (SIS), and Inter-Candidate Detection (ICD). SIS evaluates four manipulation dimensions: Authority Attribution (AA), Narrative Purposiveness (NP), Comparative Claims (CA), and Temporal Claims (TC). Evaluated on 600 Amazon product descript
As LLM-based ranking systems become ubiquitous, the incentive and opportunity for malicious manipulation, such as Generative Engine Optimization, naturally increases, necessitating defensive measures.
This research addresses a critical vulnerability in the integrity of LLM-driven information systems, impacting trust, market fairness, and the reliability of digital commerce.
The development of robust defense frameworks like SCI-Defense changes the landscape from unchecked generative manipulation to a more secure and verifiable online ecosystem.
- · Platforms using LLM-based ranking systems
- · Consumers relying on product reviews
- · Authentic businesses
- · AI security solution providers
- · Adversaries using GEO attacks
- · Companies relying on deceptive marketing
- · Bot farms
Increased integrity and trustworthiness of LLM-based ranking systems across various platforms.
A potential arms race between generative manipulation techniques and advanced defense mechanisms, driving innovation in both areas.
Enhanced regulatory scrutiny on platforms regarding their responsibility to mitigate AI-driven manipulation and ensure fair market practices.
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.LG