AI Tools & Models HOTFeaturedAll
arXiv cs.AIScore65

Can LLMs Write Backdoors? An Empirical Security Study of Weight-Space Programming

Summary

This empirical study tests whether LLMs can plant backdoors directly in neural network weights via weight-space programming and assesses the security risk.

Why this matters

Covers model and LLM · Relevant to this topic

Open original ↗

Source: arXiv cs.AI · arxiv.org

Topics

Related items