Skip to main content
Home/Risks/Marchal2024/Steganography

Steganography

Sub-category
Risk Domain

Vulnerabilities that can be exploited in AI systems, software development toolchains, and hardware, resulting in unauthorized access, data and privacy breaches, or system manipulation causing unsafe outputs or behavior.

"Steganography is the practice of hiding coded messages in GenAI model outputs, which may allow malicious actors to communicate covertly.8"(p. 9)

Part of Misuse tactics to compromise GenAI systems (Model integrity)

Other risks from Marchal2024 (22)