CyberdeceptionAtlas · Question → testES

Library / Paper

Paper

Let Them Steal: Trapping Large Language Model Extraction Attacks with Knowledge Honeypot

According to the authors, this paper proposes 'Knowledge Trap', a defense against large language model extraction attacks that redirects attackers toward low-transferability knowledge using a honeypot-style knowledge graph and breadcrumb-guided exploration, rather than blocking queries or perturbing outputs. The authors report experiments in medical and financial domains showing reduced surrogate agreement without degrading legitimate-user performance.

HoneypotApplication

What we know

According to the authors, this paper proposes 'Knowledge Trap', a defense against large language model extraction attacks that redirects attackers toward low-transferability knowledge using a honeypot-style knowledge graph and breadcrumb-guided exploration, rather than blocking queries or perturbing outputs. The authors report experiments in medical and financial domains showing reduced surrogate agreement without degrading legitimate-user performance.

Evidence limits

This entry is based solely on the published title and abstract; the full text was not reviewed and the reported effectiveness has not been independently validated.

Reviewed: author abstract. Full-text results were not analyzed.

Sources and provenance

  1. Let Them Steal: Trapping Large Language Model Extraction Attacks with Knowledge Honeypot
    abstract · 2026-10-04

Reviewed: 2026-10-04. This record may change when new evidence is found.

RIS · BibTeX