Researcher found 26,400 working cloud and AI API keys, including DeepSeek and OpenAI keys, exposed on Gitee
An independent researcher, working with MongoDB's Kingfisher team, scanned 7.25 million public repositories on China's Gitee platform. The scan found 26,400 cloud and AI keys that authenticated during testing. These included 3,170 DeepSeek keys, 1,348 OpenAI keys and an organization-admin credential for an unnamed enterprise AI platform. The scan did not establish attacker use. The researcher notified major affected vendors, but individual developers were not contacted and revocation remained unconfirmed.
First reported October 28, 2026 · Added to the register November 7, 2026 · 1 source
- Country
- China
- When it happened
- Not stated in the sources
- First reported
- October 28, 2026
What this means for you
Could this affect you?
If you or your suppliers keep code on Gitee. Developers, contractors and businesses whose Gitee repositories, including old commits, forks and mirrors, hold cloud or AI keys were exposed, and companies using China-based suppliers may be exposed even if their own code is hosted elsewhere.
What to check
- Scan full repository histories with more than one scanner.
- Revoke and replace exposed AI and cloud keys at the provider.
- Review usage, charges and API logs for those keys.
- Ask suppliers for evidence of repository scans, revoked keys and activity reviews.
Areas of your AI programme this touches
Every fact and its source (10)
- Findings publishedOct. 28“Independent researcher Ben Zimmermann published the findings Oct. 28”[1]
- Working keys found26,400 cloud and AI keys“a scan uncovered 26,400 cloud and AI keys that authenticated during testing”[1]
- Repositories scanned7.25 million public Gitee repositories“Independent researcher Ben Zimmermann published the findings Oct. 28 after examining 7.25 million public repositories”[1]
- Admin credential exposedOrganization-admin credential for an unnamed enterprise AI platform“The discoveries included an organization-admin credential for an unnamed enterprise AI platform.”[1]
- Misuse statusAttacker use not established“The scan did not establish attacker use or continued validity today.”[1]
- ResponseMajor vendors notified; developers not contacted; revocation unconfirmed“Zimmermann notified major affected vendors, but individual developers were not contacted, and revocation remained unconfirmed.”[1]
- Keys from 2025 commits12,242“Nearly half the findings, 12,242 keys, came from 2025 commits.”[1]
- DeepSeek keys verified3,170“DeepSeek3,170Zhipu”[1]
- Research partnerMongoDB's Kingfisher team“In research with MongoDB's Kingfisher team”[1]
- OpenAI keys verified1,348“OpenAI1,348The discoveries”[1]
Sources
- 26,400 Valid Cloud and AI Keys Exposed on Giteeesecurityplanet.com · October 30, 2026
How this record is classified. Severity S (3/5): real harm to one organisation, data exposed, or a containment failure. OECD level: hazard, an event that could plausibly have led to harm. Evidence: Reported, meaning press reporting only.
The facts in At a glance and in Every fact and its source are quoted from the sources listed. The summary, What the AI did and What this means for you are written from those sources and checked against them automatically before publication. Records reflect what has been disclosed, not everything that has happened. Incidents are found through news feeds and the GDELT Project.
