AEGLIX News
Researchers at Stony Brook University deployed 300 honeypot container images containing planted AWS, OpenAI, and Anthropic credentials across eight public registries over a four-month period to measure how quickly exposed secrets are discovered and abused, finding that automated scanners extracted and used the keys within seconds of publication on Docker Hub.
The honeypot images were pulled more than 75,000 times, and the embedded AWS keys alone generated over 140,000 API calls originating from 5,000 unique IP addresses. Docker Hub accounted for nearly all observed AWS key activity, while the Harbor registry also saw some abuse. The researchers labeled the automated scanners 'ContainerBots' as a loose umbrella term, noting the scans showed no evidence of coordination by the same operators.
The fastest observed use of a planted AWS key occurred 7.86 seconds after the image appeared on Docker Hub. The median time to the first API call was 51.19 seconds, and all 100 AWS keys were used within 15 minutes. Exposed OpenAI and Anthropic keys were similarly hijacked to exhaust test-account balances. Monitoring rarely detected anyone actually running the honeypot images, leading researchers to conclude that scanners extracted credentials by inspecting image layers directly rather than executing the containers.
The researchers recommend that Docker Hub scan images for secrets before pushing them live and warn publishers when secrets appear embedded. They also suggest restricting or rate-limiting access to Docker Hub's public feed of recently published or updated images to thwart mass-scanning tools. The findings were reported to Docker in May along with the full paper, but the team received only an acknowledgment with no follow-up replies. A small-scale August experiment showed no changes had been implemented on the platform.
Report prepared 2026-11-17 · Based on reporting from BleepingComputer. Source publication date above; subsequent developments may not be reflected.
Sources
Read original source · BleepingComputer (opens in a new tab)