Cloudflare Discloses Tenant Isolation Failure in GLM-5.3 Model Affecting 155 Requests
Sign in✓ Signed in
Cloudflare disclosed on October 26 that its GLM-5.3 Workers AI model exposed customer data across user sessions for 155 responses between August 27 and September 9, 2026, caused by a configuration mismatch and an SGLang inference engine bug. The leak affected only long-context requests exceeding 681,280 tokens, the capacity limit of the key-value cache pool on Cloudflare's GPU clusters.
Continue reading
The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in✓ Signed in — this article isn’t included in your current plan.
Unlocking the full article...