Docker Hub secrets stolen within seconds by automated scanners
Researchers found that exposed credentials in public container images are extracted and exploited by automated bots within seconds of publication.
Stony Brook University researchers demonstrated that credentials exposed in public Docker Hub images are extracted and exploited within as little as eight seconds. By publishing 300 honeypot images across eight registries over four months, the team observed that automated scanners, dubbed ContainerBots, pulled these images more than 75,000 times. The fastest observed use of a planted AWS key occurred 7.86 seconds after publication, with the median time for the first API call being 51.19 seconds. All 100 AWS keys were used within 15 minutes, receiving over 140,000 API calls from 5,000 unique IP addresses. Exposed OpenAI and Anthropic keys were also quickly hijacked to exhaust test-account balances.
The researchers concluded that these scanners generally extract credentials by inspecting image layers directly without instantiating the container. Monitoring rarely detected anyone actually running the honeypot images. Docker Hub accounted for nearly all AWS activity observed, though the Harbor registry also saw some abuse. The team presented these findings at the ACM Conference on Computer and Communications Security.
Researchers recommend that Docker Hub scans images for secrets before pushing them live and warns publishers if secrets appear. They also suggest restricting or rate-limiting access to the public feed of recently published images. Despite reporting findings to Docker in May, the researchers received no follow-up replies other than an acknowledgment. A small-scale experiment run in August showed no changes regarding defenses implemented on the platform.
Sources
Every sentence of this story was checked against facts quoted from these articles. How we make the news
Spotted an error? Tell us
