Preliminary research

Cache Me, Catch You: Exploiting LLM Caching Layers in vLLM, GPTCache & Friends

Xiangfan Wu, Lingyun Ying, Haipeng Qu, Guoqiang Chen and Yacong Gu

AI-collected research leads through 6 October 2026, including bounded month-by-month reviews of selected social and community sources from January through September. Unranked, incomplete, not community-vetted, and subject to change.

Research on LLM serving caches (vLLM, GPTCache and peers) where the layer deciding whether two requests are 'the same' is fooled. All three cache types rest on serialize-key-reuse, so an attacker crafts colliding inputs: hash-colliding padding poisons a shared system-prompt or block-wise KV entry so a malicious block is skipped; near-neighbour embeddings make semantic and RAG caches serve a planted answer; byte-only image hashing makes moderation reuse a benign verdict. Three CVEs resulted.

Record

Researcher

Xiangfan Wu, Lingyun Ying, Haipeng Qu, Guoqiang Chen and Yacong Gu

Format

Whitepaper

In the archive

• Collected in the 2026 preliminary research sweep

Related sources

• Paper (NDSS 2026)
• Code
• Kv Cache Collision advisory
• Image Hash Collision (tobytes) advisory
• PNG tRNS Transparency Bypass advisory

Tags

• llm
• cache-poisoning
• hash-collision
• rag
• prompt-injection
• ai-agent
• owasp-a02-2021
• owasp-a03-2021

This page is the archive's own catalogue record. The research is the work of Xiangfan Wu, Lingyun Ying, Haipeng Qu, Guoqiang Chen and Yacong Gu, first published at the original source. Preserved copies are kept so the citation survives its host.