Cloudflare discloses cross-customer data exposure on Workers AI, traced to an SGLang bug
Cloudflare says 155 responses from the GLM-5.3 model on Workers AI between Aug 27 and Sep 9 contained data from other customers' requests, all for prompts over 681,280 tokens. A mismatch between the context window and the KV cache pool triggered a bug in SGLang's disaggregated prefill/decode mode. Cloudflare says it found no evidence of deliberate exploitation; SGLang published a coordinated fix.
Why it matters
Engines often reuse KV cache pages across requests; an ownership bug in disaggregated serving can leak one user's data into another's reply. Cloudflare urges SGLang operators to match KV pool size to context length and upgrade to v0.5.22+.
Sources3
CloudflareOfficial · BlogCross-request KV cache exposure in the GLM-5.3 Workers AI model, and our remediation with SGLangView original · LMSYS OrgMediaSGLang security update: improving request isolationView original ·
- @lmsysorgXSGLang security update: improving request isolation. Joint work published in coordination with Cloudflare, this post covers the KV cache issue in disaggregated serving, the fixes, and recommended actions for operators: lmsys.org/blog/2026-10-2…View original ·
Written by dalao.ai: the title, summary and why it matters are our own words. The original reports are the record; their publishers keep authorship.
