Devs say Chinese AI company silently uploaded hundreds of megabytes of local workspace data, company apologizes — Z.AI, the firm behind the GLM models, didn’t ask for user consent and made 564 attempts to exfiltrate 313MB archive
- ID
- 26778
- Status
- summarized
- Published
- 21 Sep 2026, 7:59 PM
- Fetched
- 21 Sep 2026, 8:53 PM
- Provider
- Tom's Hardware
- Category
- technology
- Original URL
- https://www.tomshardware.com/tech-industry/artificial-intelligence/devs-say-chinese-ai-company-silently-uploaded-hundreds-of-megabytes-of-local-workspace-data-z-ai-the-firm-behind-the-glm-models-didnt-ask-for-user-consent-and-made-564-attempts-to-exfiltrate-313mb-archive
- Source URL
- https://www.tomshardware.com/feeds/all
Summary
- Score
- 7.5
- Created
- 21 Sep 2026, 8:55 PM
- Tags
- Audience
- developersvibe_codersai_agent_usersai_ml_learners
What happened
Z.AI, the company behind the GLM models, allegedly made 564 attempts to silently exfiltrate a 313MB archive of local workspace data without user consent. The company has since apologized. The incident raises serious trust questions for developers using GLM-based tools or coding assistants.
Why it matters
If you are evaluating or using GLM models or any Z.AI tooling in your dev environment, treat them as untrusted: run them in containers or VMs, restrict network egress, and audit what local files they can access. This is a concrete data-exfiltration incident, not a hypothetical privacy concern — 564 upload attempts against a 313MB archive means the tool was actively scanning and packaging workspace contents.
Discussion angle
What's your baseline isolation strategy for AI coding assistants and local LLM tooling — do you containerize everything, or trust the vendor? This incident is a good prompt to compare sandboxing approaches (Docker, devcontainers, firejail, network policies) before the next tool gets caught doing the same thing.