Browsable glossary
The AI Governance Glossary
Plain-English definitions for the terms that come up in every AI transformation conversation
Back to resources
GlossaryFree to read
No jargon and no unnecessary legal language — just clear, concise definitions for the terms you will hear in leadership meetings, vendor conversations, transformation programmes, and compliance reviews.
21 terms
A
- AI Advisor
- A guided, conversational experience that asks structured questions about an organisation and produces a tailored recommendation — for example, its likely AI maturity stage, key priorities, or a suitable product configuration.
- AI Governance
- The policies, processes, controls, and roles that determine how AI is approved, used, monitored, and reviewed within an organisation. It is the operational counterpart to an AI strategy.
- AI Literacy
- The practical knowledge and skills employees need to use AI safely, responsibly, and effectively. It is distinct from technical AI expertise and relevant across roles and departments.
- AI Maturity
- A measure of how structured, embedded, governed, and value-oriented an organisation’s AI capability has become. It is commonly assessed across areas such as strategy, workforce enablement, governance, and measurable business value.
- AI Readiness
- An organisation’s current level of preparedness to adopt and scale AI responsibly, including its skills, data foundations, governance, leadership alignment, and operating model.
B
- Bias in AI
- Systematic patterns in an AI system’s outputs that may unfairly favour or disadvantage certain people or groups. Bias can originate from training data, system design, implementation choices, or the context in which an AI system is used.
C
- Compliance by design
- An approach in which regulatory, policy, privacy, security, and control requirements are incorporated into a system or process from the beginning rather than checked and added afterwards.
E
- EU AI Act
- The European Union regulation governing the development, provision, deployment, and use of AI systems. It applies different requirements depending on the role of the organisation, the type of AI system, and its risk classification.
- Explainable AI (XAI)
- Methods and practices that help people understand how an AI system reached an output, recommendation, or decision. Explainability is particularly important where results must be reviewed, challenged, justified, or audited.
G
- General-Purpose AI (GPAI)
- An AI model designed to perform a broad range of tasks rather than one narrowly defined function. Large language models are a common example.
- Governance maturity
- The degree to which an organisation’s AI governance processes are consistently understood, followed, evidenced, reviewed, and improved in day-to-day operations.
H
- Hallucination
- An output generated by an AI system that appears plausible but contains incorrect, unsupported, or fabricated information. Hallucinations are an important reason why human review remains necessary for consequential uses of AI.
- High-risk AI system
- An AI system used in a context where errors may significantly affect health, safety, or fundamental rights and which may therefore be subject to stricter requirements under the EU AI Act.
- Human in the loop
- A design and governance principle in which a person reviews, can override, or must approve an AI-generated output before it takes effect.
- Human oversight
- The measures that enable people to understand, supervise, challenge, intervene in, or stop the operation of an AI system where necessary.
M
- Model risk
- The risk that an AI model produces unreliable, incorrect, biased, insecure, or otherwise unsuitable outputs and that those outputs are used without adequate controls or scrutiny.
P
- Prompt engineering
- The practice of structuring instructions, context, constraints, and examples so that an AI system produces more accurate, useful, consistent, and policy-aligned outputs.
R
- Responsible AI
- An umbrella term for the principles and practices that support the safe and trustworthy development and use of AI, including fairness, transparency, accountability, privacy, security, safety, and human oversight.
S
- Shadow AI
- The use of AI tools within an organisation without sufficient visibility, approval, or control from leadership, IT, security, legal, or compliance teams. A common example is employees using personal AI accounts for work-related tasks.
T
- Tool sprawl
- The uncontrolled growth of overlapping and disconnected AI tools across an organisation without a shared strategy, ownership model, integration approach, or governance framework.
U
- Use Case Register
- A maintained and structured record of an organisation’s AI use cases, including their purpose, owner, users, affected processes, data, risk classification, approval status, controls, and review history.
Continue exploring
Other free, plain-English resources from casaios.