LATIDIA · Ciberseguridad

SRAF: Huella digital adversaria sigilosa y robusta para la verificación de derechos de autor de modelos de lenguaje grandes

arXiv: 2505.06304v5Tipo de anuncio: reemplazar Resumen: La protección de la propiedad intelectual (IP) para modelos de lenguaje grandes (LLM) se ha convertido en una preocupación crítica a medida que aumentan el robo de modelos y la comercialización no autorizada.

La noticia

arXiv:2505.06304v5 Announce Type: replace Abstract: The protection of Intellectual Property (IP) for Large Language Models (LLMs) has become a critical concern as model theft and unauthorized commercialization escalate. While adversarial fingerprinting offers a promising black-box solution for ownership verification, existing methods suffer from significant limitations: they are fragile against downstream model modifications, sensitive to system prompt variations, and easily detectable due to high-perplexity input patterns. In this paper, we propose \textbf{SRAF}, a stealthy and robust adversarial fingerprinting framework. SRAF employs a synergistic joint optimization strategy across homologous model variants and diverse chat templates, forcing the fingerprint to anchor onto the invariant intrinsic comprehension features of the model family. Furthermore, we