PulseAugur
EN
LIVE 01:32:13

Cloudflare GLM-5.3 model exposed user data via cache bug

By PulseAugur Editorial · [1 sources] ·

Cloudflare reported a security vulnerability in its GLM-5.3 model where a bug in SGLang allowed requests to access leftover memory from previous sessions. This issue, which occurred between August 27 and September 9, affected users sending prompts up to 700k tokens, with the cache holding 681k tokens. Cloudflare has since released an update to address the cache size and context length mismatch. AI

IMPACT Highlights potential security risks in LLM infrastructure and the importance of proper cache management.

RANK_REASON A specific bug in an AI model's infrastructure led to a data exposure incident.

Read on Mastodon — mastodon.social →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

Cloudflare GLM-5.3 model exposed user data via cache bug

How we ranked this

Signal score
2 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
A specific bug in an AI model's infrastructure led to a data exposure incident.
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
safety, infra
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
Same-day
Cluster formed today. Ranking reflects the current source set at time of score.

Full methodology in our editorial standards.

COVERAGE [1]

  1. Mastodon — mastodon.social TIER_1 English(EN) · mohith808 ·

    POV: you send a 700k token prompt to an AI and the model reads leftover memory from someone else's request cloudflare says that hit 155 responses on its GLM-5.3

    POV: you send a 700k token prompt to an AI and the model reads leftover memory from someone else's request cloudflare says that hit 155 responses on its GLM-5.3 model, aug 27 to sep 9. the context window was 1M tokens but the cache only held 681k, and an SGLang bug let rejected r…

    LINKS blog.cloudflare.com/kv-cache-exposure-rem…

RELATED ENTITIES

RELATED TOPICS