PulseAugur
EN
LIVE 11:58:19

Cloudflare GLM-5.3 model leaked customer data due to KV cache flaw

By PulseAugur Editorial · [1 sources] ·

Cloudflare experienced a security incident where a tenant isolation flaw in its GLM-5.3 model, hosted on Workers AI, allowed prompts longer than 681,280 tokens to return data from other customers' sessions. The issue stemmed from a configuration mismatch in the KV cache sizing within the SGLang inference engine, which was not properly handling prompts exceeding the allocated memory. Cloudflare identified and fixed the root cause within hours of the incident being declared, and a public postmortem was released six weeks later after upstream fixes were deployed. AI

IMPACT Highlights the critical need for robust tenant isolation in hosted LLM inference to prevent data leakage.

RANK_REASON The article details a specific security vulnerability and its resolution within a hosted AI model service, which falls under the category of tooling and infrastructure.

Read on dev.to — LLM tag →

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

Cloudflare GLM-5.3 model leaked customer data due to KV cache flaw

How we ranked this

Signal score
33 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
The article details a specific security vulnerability and its resolution within a hosted AI model service, which falls under the category of tooling and infrastructure.
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
product, safety
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
Breaking (< 6h)
Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

Full methodology in our editorial standards.

COVERAGE [1]

  1. dev.to — LLM tag TIER_1 English(EN) · Damiano ·

    One Token Leaked 155 AI Responses: Inside Cloudflare's GLM-5.3 KV Cache Exposure

    <p>On September 8, 2026, a security researcher told Cloudflare that very long prompts sent to one of its hosted models sometimes came back with text from other customers' sessions. Cloudflare reproduced the leak within a day, took the model offline forty minutes after declaring a…

RELATED ENTITIES

RELATED TOPICS