Stony Brook study: leaked AWS keys in Docker Hub images used within 8 seconds
Researchers Billy Tsouvalas and Nick Nikiforakis of Stony Brook University published 300 honeypot container images with planted AWS, OpenAI and Anthropic credentials across eight public registries over four months. The images were pulled more than 75,000 times, and the fake AWS keys drew over 140,000 API calls from roughly 5,000 unique IP addresses, with the fastest key abuse occurring under eight seconds after publication and all 100 keys used within 15 minutes. The findings were presented at the ACM CCS conference.
GoKawiil's interpretation of the reporting above, not reported fact.
The speed and scale of key harvesting suggest automated scanners continuously monitor public registries like Docker Hub for exposed secrets, leaving almost no window for developers to catch and revoke leaked credentials manually. Because the attacking traffic came from thousands of unrelated IPs, the researchers suggest this is driven by many independent opportunistic scanners rather than one coordinated actor, implying the problem is widespread rather than isolated. This could push cloud providers and registry operators toward automated secret-scanning and near-instant key revocation as a baseline defense.
- Planted AWS keys in Docker Hub images were abused in as little as 7.86 seconds after publication
- 75,000+ pulls and 140,000+ API calls came from about 5,000 unique IPs during the 4-month study
- Exposed OpenAI and Anthropic keys were quickly drained, highlighting risk beyond cloud infrastructure credentials
Source: bleepingcomputer.com, 2026-11-17
Published there as: “Cloud keys leaked in Docker Hub images are stolen within seconds”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.