Glossary
Algorithmic Accountability
The principle that named individuals or roles are responsible for the outcomes of AI systems, including the decisions made, the controls applied and the evidence retained.
Term: Algorithmic Accountability
Algorithmic accountability is the principle that named individuals or roles are responsible for the outcomes of AI systems, including the decisions made, the controls applied and the evidence retained.
Why it matters
Governance frameworks describe what good looks like; accountability decides who answers for it. Without named owners, AI governance frameworks such as ISO 42001 and the EU AI Act stay aspirational. A control with no owner is a control no one maintains, and evidence with no owner is evidence no one can produce when a regulator, customer or auditor asks.
For UK organisations in regulated markets, this is a structural requirement. When a UK regulator or customer challenges an AI decision, the first question is always who owns the response.
How it works in practice
We treat algorithmic accountability as a precondition for every in-scope AI system. Each system carries a named control owner and a named evidence owner against all six dimensions of the CIA+EFT Framework: confidentiality, integrity, availability, explainability, fairness and traceability. The control owner maintains the safeguard; the evidence owner produces the proof on demand. No dimension is left unassigned.
Book an AI Security Programme scoping call with QL Security.
Related terms: AI Governance, CIA+EFT Framework and ISO 42001.
Want this in context?
See how this term fits into the broader programme of work.