Security · Updated 27 Oct, 03:31 am IST
Cloudflare fixes cross-request data leak in GLM-5.3 Workers AI model
Why it matters for readers: You should know that a small number of long AI requests briefly exposed other customers' data and the provider has fixed it.
- A tenant-isolation bug in the GLM-5.3 Workers AI model caused some responses to include data from other customers' recent requests.1
- Cloudflare identified 155 affected responses, all to prompts longer than 681,280 tokens, occurring between August 27 and September 9, 2026.1
- The issue was reported by security researcher Gadi Evron on September 8, 2026; Cloudflare remediated the bug and restored the model on September 9.1
- Cloudflare confirmed no other Workers AI models were affected and found no evidence of deliberate exploitation.1
- Cloudflare published a coordinated disclosure with SGLang, the open-source inference engine used in production, and is working with upstream maintainers to harden deployments.1
Get a brief like this every morning
Uzha reads hundreds of sources and gives you the stories that matter for your work, with every source linked. Free.
Get started