Docker Images: Cloud Keys Stolen Within Seconds
Stony Brook University researchers Billy Tsouvalas and Nick Nikiforakis conducted a four-month study placing honeypot images containing cloud credentials across eight public container registries. They embedded AWS, OpenAI, and Anthropic API keys along with decoy SSH and database credentials to track when attackers would discover and exploit them.
The scale of exploitation was substantial. The 300 images were pulled over 75,000 times, with the planted AWS keys generating more than 140,000 API calls from 5,000 unique IP addresses. The speed was alarming: the fastest exploitation occurred 7.86 seconds after publication on Docker Hub, while all 100 AWS keys were accessed within 15 minutes. The median time between publishing and first API call was 51.19 seconds. Docker Hub accounted for nearly all the AWS activity, though Harbor also showed some key abuse. OpenAI and Anthropic credentials were quickly exploited to drain test-account balances.
Analysis revealed attackers use automated scanners—dubbed “ContainerBots” by the researchers—that extract secrets through static analysis of image layers rather than by running containers. The researchers presented their findings at the ACM Conference on Computer and Communications Security and recommended Docker implement pre-publication secret scanning and restrict access to feeds of newly published images. When they reported findings to Docker in May, they received only an acknowledgment with no follow-up response.
Read the full article
Continue reading at BleepingComputer →This is an AI-generated summary. Read the original for the full story.