We found 4,972 "hello" messages hiding thousands of malicious commands targeting Flowise
Era researchers discovered Greeter, an attack framework targeting Flowise, an AI development platform. They observed 4,972 malicious JavaScript messages via honeypots, which performed reconnaissance, searched for credentials, probed cloud metadata, and prepared for persistence. The malicious code cleverly avoided being saved in the workflow. The team identified thousands of exposed Flowise instances, including vulnerable, unauthenticated ones. This attack highlights the significant risk of compromised AI applications serving as entry points into broader cloud environments.