Pillar two · Accountability

Who is answerable for this system?

The test is simple and unforgiving. Pick any AI system you run. Can you name its Business, Technical, Model and Data Owner in sixty seconds, without asking anyone? If not, you do not have an accountability model — you have a diagram of one.

11 weighted checks. Nothing you enter leaves your browser.

The named roles

Business OwnerFirst line
Business value, intended use, acceptable-use boundaries, and the continue, change or retire decision.
Technical OwnerFirst line
Engineering integrity in operation, incidents, change control, technical evidence. Holds authority to pause without convening anyone.
Model or Agent OwnerFirst line
Model selection, evaluation, fine-tuning, prompts, tool permissions, memory scope and content filters.
Data OwnerFirst line
Dataset quality, provenance, licensing, consent, retention and right to use.
Risk and Compliance
Ongoing conformance to policy and obligation, and the triggers that reopen assessment.
Privacy Officer / DPO
Lawful basis, impact assessments, data-subject rights, cross-border transfer.
Security Officer
Threat model, vulnerability management, access control, non-human identity, logging.
Legal Counsel
Contractual exposure, intellectual property, liability, regulatory interpretation.
Independent Review
Second-line challenge and assurance. Not the same people who built it.
Executive Sponsor
Ultimate accountability for high-risk systems and for the portfolio as a whole.

The checks

  1. Named owners — Can you name the Business Owner — an individual, not a team?
  2. Named owners — Can you name the Technical Owner?
  3. Named owners — Can you name the Model or Agent Owner?
  4. Named owners — Can you name the Data Owner?
  5. Authority — Can the Technical Owner pause the system without convening a committee?
  6. Authority — Is there a named individual who has accepted the residual risk?
  7. Evidence — Are decisions held as records rather than in email threads and chat?
  8. Evidence — Is the periodic review in date, with owners re-attested?
  9. Evidence — Is there evidence a human has ever disagreed with the system?
  10. Third parties — For third-party AI, is the provider and deployer split documented?
  11. Third parties — Would you find out if the vendor silently changed the model underneath you?

Bands

Defensible 90%+
You could answer a regulator, an auditor or a board in minutes. Keep the attestations current.
Mostly held 70%+
The structure is real but has soft spots. Close them before you need them rather than during an incident.
Thin 45%+
Accountability exists on paper and in a few people’s heads. It will not survive a departure or an incident.
Absent 0%+
There is no answer to "who is accountable for this". Everything downstream — controls, evidence, approvals — rests on nothing.

Three lines

First line —
Business and technology delivery — the four owner roles. Owns the system, the risk and the day-to-day decisions.
Second line —
Risk, compliance, privacy, security, responsible AI. Sets policy, challenges the first line, reports independently.
Third line —
Internal audit. Provides independent assurance that the first two lines work as described.

Moments that require a name

  • — At intake, when owners are first named
  • — At approval, when conditions are accepted
  • — At go-live, when the owner formally accepts the system
  • — At material change, including a vendor changing the model underneath you
  • — At incident, when someone must have standing authority to stop it
  • — At periodic review, when owners re-attest
  • — At retirement, when someone confirms the data is handled correctly

Common questions

Who should be accountable for an AI system?
Four named individuals, not teams. A business owner accountable for value, intended use and the continue-change-retire decision. A technical owner accountable for operational integrity, incidents and change control, holding standing authority to pause without convening anyone. A model or agent owner accountable for model selection, evaluation, prompts, tool permissions and memory scope. A data owner accountable for provenance, licensing, consent and right to use.
What is the difference between accountability and responsibility for AI?
Responsibility is shared and can be distributed across a team. Accountability sits with one person and cannot be divided without disappearing. It is also distinct from blame, which is punitive and retrospective — accountability is the authority to act. Risk accepted by a committee is risk accepted by nobody.
Who is accountable for a third-party AI system we did not build?
Accountability must still terminate inside your organisation. The provider carries obligations for the system they place on the market; as the deployer you remain accountable for how it is used, the data you put into it, the oversight around it and the outcomes it produces. That split should be documented in the contract rather than assumed — and putting your own name on someone else’s system can make you the provider.
How do you prove AI accountability to an auditor?
Named individuals recorded against each role with the date they accepted it; decisions, approvals, conditions and overrides held as immutable records rather than email; attestations that are in date; standing pause authority documented before an incident rather than during one; and evidence that a human has actually overridden the system at least once. Oversight that leaves no trace of disagreement is decorative.