Security Assessment of DeepSeek Harness with A.I.G: Evaluating Resistance to Indirect Prompt Injection
Researchers evaluate indirect prompt injection risks in DeepSeek Harness using controlled taint and dual judges, finding notable success rates across text and file channels and rec…
Hugging Face · Daily Papers
·Zonghao Ying, Xiangfan Wu
·
·▲ 4 upvotes
Este artigo está em destaque na seleção diária de papers do Hugging Face, curada pela comunidade de pesquisa em IA.
Autores: Zonghao Ying, Xiangfan Wu, Huiyu Wu, Xing Zheng, Huangsheng Cheng, Xiaorong Shi
- 4 upvotes da comunidade
- Temas: indirect prompt injection, DeepSeek Harness, AI-Infra-Guard, agent loop, tool registry, model adapter
Resumo
Resumo original (em inglês), extraído do paper:
Researchers evaluate indirect prompt injection risks in DeepSeek Harness using controlled taint and dual judges, finding notable success rates across text and file channels and recommending controls between untrusted content and sensitive actions.