ISACA's AAISM credential is designed for security professionals managing complex AI systems. it combines risk management and technical controls.
Answer : A security vulnerability where an attacker manipulates an LLM's output by injecting instructions that override its original system prompt.
Can be used to bypass safety filters or extract sensitive configuration details.
Answer : Using a large number of queries to a target model to build a clone or approximation of its logic and intellectual property.
Allows competitors to steal expensive proprietary AI logic for the cost of API calls.
Answer : It adds mathematical noise to data before training to ensure individual data points cannot be identified, while preserving group patterns.
The gold standard for protecting privacy in high-scale machine learning datasets.
Answer : Modifying an input (like an image or email) so subtly that it is invisible to humans but causes the AI to misclassify it.
Used to bypass spam filters or fool facial recognition and autonomous driving systems.
Answer : Expanding traditional security threat models to include AI-specific attack vectors like training data poisoning and membership inference.
Essential for building resilient AI architectures by design.