Credentials and access secrets embedded in public Docker Hub images are being compromised by attackers extremely quickly. A study by Stony Brook University researchers Billy Tsouvalas and Nick Nikiforakis revealed that exploitation can occur within eight seconds of an image being published. The median time for the first API call to a leaked AWS key was 51.19 seconds, with all 100 test AWS keys used within 15 minutes.
The researchers conducted an experiment over four months, deploying 300 honeypot images across eight public container registries. These images contained decoy AWS, OpenAI, and Anthropic API keys, as well as SSH and database credentials. They monitored the usage of these keys and detected when images were built, run, or had a shell opened. The findings were presented at the ACM Conference on Computer and Communications Security (CCS).
The honeypot images were pulled over 75,000 times. Planted AWS keys generated more than 140,000 API calls from 5,000 unique IP addresses. This traffic included checks for key validity, resource listing, and resource consumption. OpenAI and Anthropic keys were quickly used to deplete test account balances. Docker Hub accounted for most of the AWS activity, with some abuse also noted on Harbor.
The researchers concluded that attackers, dubbed 'ContainerBots,' primarily extract credentials through static analysis of image layers rather than by running the containers. Monitoring rarely showed instances of the honeypot images being run, indicating that scanners unpack image layers or query registry APIs to access filesystem contents directly. This method allows for rapid discovery and exploitation of exposed secrets.
Given the prevalence of exposed secrets in real-world images and the effort required to revoke them, the researchers recommend that Docker Hub implement scanning for secrets before images go live. They suggest that publishers should be warned if their images appear to contain embedded credentials, to prevent such rapid compromise.
✨ This summary was generated by AI from the outlets' reporting listed below. It is not independently verified and may contain errors — check the original sources. How BrevFeed works →
One email each morning: the day's tech stories, clustered across outlets and summarized. No account needed.
One email a day. Unsubscribe in one click, any time.
Spend a few minutes, get the whole day. Every topic's top stories in one hands-free rundown — listen, watch, or read the transcript.
▶ Play today's briefNew every morning, and the back catalogue is archived by date.
Researchers found that credentials exposed in public Docker Hub images are extracted and used by attackers in as little as eight seconds. An experiment involving 300 honeypot images across eight registries showed rapid exploitation of AWS, OpenAI, and Anthropic API keys, primarily through static analysis of image layers.