HCIIF™INFORMATION INTEGRITY

HCIIF & AI

Confidence before
prediction, generation or action.

HCIIF strengthens the information foundations that predictive, generative and agentic AI rely upon—without claiming to govern the complete AI system.

ONE INFORMATION-ASSURANCE FOUNDATION

Three AI modes.
One material question.

Is the information sufficiently trusted for the intended purpose, risk and consequence?

Before AI predicts, generates or acts, organisations need to understand what confidence the underlying information justifies.

Home/HCIIF & AI

Model-neutral.
Reliance-specific.

HCIIF applies wherever information is used to support a judgement, generate an output or initiate an action.

Each AI mode raises a different reliance question. Their relative significance will continue to be explored as AI adoption, organisational use and associated consequences develop.

01

Predictive AI

Can we rely on this prediction?

Tests historical information, features, assumptions and changing conditions against the intended decision.
02

Generative AI

Can we rely on this generated information?

Tests provenance, transformation, confidence, caveats and suitability for the intended use.
03

Agentic AI

Can this system rely on the information sufficiently to decide or act?

Connects confidence conditions with authority, oversight, escalation, stopping and recovery.
AGENTIC AI · FROM INFORMATION TO ACCOUNTABLE ACTION

Authority must travel
with confidence.

An agent may retrieve information, choose between options, initiate action and pass outputs between systems. Each hand-off needs an explicit confidence and authority boundary.

01Information
02Confidence conditions
03Authority
04AI action
05Human oversight
06Recall & recovery
WHAT HCIIF MAKES VISIBLE

The grounds for reliance.
Not just the output.

HCIIF keeps the evidence, limitations, permitted use and accountable authority connected to information as it moves through AI-enabled environments.

HCIIFInformation
integrity
  1. 01Enterprise Information Asset
  2. 02Provenance and transformation history
  3. 03Required and Current Confidence
  4. 04Confidence Applicability
  5. 05Rights, Caveats and permitted use
  6. 06Decision and Acceptance Authority
  7. 07Confidence Events, Drift and Debt
  8. 08Recall and recovery requirements
  9. 09Human review, escalation and stopping conditions

THE BOUNDARY

Information integrity is essential.
It is not the whole AI system.

HCIIF complements AI governance, model risk, safety, security, privacy and engineering controls. It provides disciplined evidence about whether the information entering, moving through or emerging from AI justifies reliance for a defined purpose.

HCIIF & AI

Trusted information
before consequential action.

HCIIF is AI-model-neutral, with particular relevance where AI-generated information informs or initiates consequential action.