PulseAugur
EN
LIVE 17:03:36

Cloudflare Workers AI leaks customer data due to GLM-5.3 KV cache bug

By PulseAugur Editorial · [1 sources] ·

A critical bug in Cloudflare Workers AI, affecting the GLM-5.3 model, led to the leakage of 155 customer responses between August 27 and September 9. The issue stemmed from the inference engine serving stale KV cache pages from previous requests, rather than rejecting oversized prompts. This occurred because the KV pool size was smaller than the model's context window, causing an uninitialized read of data from other customers' prompts and generations. Cloudflare has since detailed the incident and implemented fixes, including KV checksums and per-page ownership tags, to prevent recurrence. AI

IMPACT Highlights a critical data leakage vulnerability in AI inference engines, emphasizing the need for robust KV cache management and security audits.

RANK_REASON The article details a specific bug and its resolution within an AI platform, including a diagnostic script, rather than a new model release or significant industry shift.

Read on dev.to — LLM tag →

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

Cloudflare Workers AI leaks customer data due to GLM-5.3 KV cache bug

How we ranked this

Signal score
34 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
The article details a specific bug and its resolution within an AI platform, including a diagnostic script, rather than a new model release or significant industry shift.
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) · Sam LABBE ·

    155 responses leaked between AI customers: the GLM-5.3 KV cache bug — and the canary test for YOUR provider

    <blockquote> <p>Between August 27 and September 9, some responses from GLM-5.3 on Cloudflare Workers AI contained <strong>other customers' prompts and partial generations</strong>. Not a breach of a database, not a prompt injection — the inference engine itself served <strong>sta…

RELATED ENTITIES

RELATED TOPICS