Declared Intent is offered as a proposed assurance artefact for stress-testing.
It may be relevant to:
- AI assurance teams
- Service assessment functions
- Regulatory sandboxes
- Audit bodies
- Cross-government digital governance units
The Core Question
Would a structured, immutable intent declaration:
materially improve clarity, accountability, or audit outcomes in AI-enabled systems?
This proposal is an attempt to test that question.
What We Are Seeking
We are not seeking:
- endorsement
- adoption
- mandate
- funding
We are seeking:
- desk critique
- stress-testing against hypothetical or past cases
- identification of failure modes
- identification of duplication with existing artefacts
How It Could Be Tested
Examples of low-risk engagement:
- Retrospective mapping – Apply the DIR schema to a past AI-enabled service
- Service assessment walkthrough – Evaluate whether a DIR would have clarified review
- Regulatory sandbox exploration – Test whether the artefact improves supervision clarity
Alignment With Existing Frameworks
The DIR does not replace:
- DPIAs
- Model cards
- Risk registers
- ISO/IEC 42001 artefacts
- EU AI Act documentation
It aims to provide a shared reference point that those processes can evaluate.
Status
This is an early-stage proposal.
It is intentionally minimal and incomplete.
Engagement and critique are welcome.
feedback@declaredintent.org