Skip to main content
Home/Risks/IBM2025/Prompt injection attack

Prompt injection attack

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.

"A prompt injection attack forces a generative model that takes a prompt as input to produce unexpected output by manipulating the structure, instructions, or information contained in its prompt."

Supporting Evidence (1)

1.
"Injection attacks can be used to alter model behavior and benefit the attacker."

Other risks from IBM2025 (63)