NIST.AI.600-1

Generative AI (GAI)

A subclass of AI models that emulate input data structures to create derived synthetic content such as text, images, and audio.

NIST AI Risk Management Framework (AI RMF)

A framework intended for voluntary use to improve organizations' ability to integrate trustworthiness into AI product design, development, and evaluation.

Confabulation

The production of confidently stated but erroneous or false content by AI systems, potentially misleading users.

Data Privacy

The risks connected to unauthorized access, usage, or disclosure of personal or sensitive information, particularly when used in training AI models.

Harmful Bias

The amplification or perpetuation of historical, societal, or systemic biases in AI outputs due to biased training data.

Algorithmic Monocultures

A phenomenon where repeated use of the same algorithms leads to increased vulnerability and risk of correlated failures.

Deepfake

A form of synthetic media in which a person in an existing image or video is replaced with someone else's likeness.

Human-AI Configuration

The arrangements and interactions between humans and AI systems that can influence the effectiveness and outcomes of AI applications.

Information Integrity

The quality and reliability of information as it pertains to being truthful, accurate, and verifiable within a context.

Provenance Data Tracking

Methods used to trace the origin and history of content created by AI systems, ensuring authenticity and integrity.

Environment Impact

The potential adverse effects that AI systems may have on the environment, particularly through resource consumption.

Incident Disclosure

The process of documenting and reporting occurrences where AI systems contribute to harms or failures, aiming to foster transparency.