Agentic data science with the Data Agent Kit
cloud.google.com·7h ago
AI Research
arXiv CS.AI·Sep 4·5 min read
When Optimization Becomes Manipulation: Defending Generative Search against Malicious Generative Engine Optimization
TL;DR
GEO Defender is a novel defense mechanism designed to protect generative search engines from malicious Generative Engine Optimization (GEO) attacks. It significantly reduces attack success rates from over 50% to just 6.2% while maintaining the quality of benign content.
✦ Why It Matters
Implement GEO Defender to enhance the resilience of your generative search applications against manipulation.
Key Takeaways
- Integrate GEO Defender to protect against malicious content manipulation.
- Utilize Shield Reranker to maintain relevance while filtering out harmful documents.
- Leverage TFSG for improved guidance on source usage in language models.
- Monitor and evaluate the effectiveness of defenses against new GEO attacks.
How It Works
GEO Defender operates in two stages: first, Shield Reranker assesses and demotes documents altered by GEO while ensuring that relevant content remains accessible. Second, TFSG compiles successful defense strategies into a natural-language library, which informs the language model's decision-making during answer generation.