Northern Ireland’s Draft Artificial Intelligence Strategy is primarily aimed at the public sector, but one section is particularly relevant to businesses supplying AI systems or AI-enabled services.

The draft says its principles would be applied through procurement, with suppliers expected to evidence how their AI systems support requirements around human oversight, fairness, transparency, accountability, sustainability, training and societal benefit.

That changes the conversation. For the last few years, businesses have largely been encouraged to adopt AI. Find efficiencies. Automate work, experiment and move faster.

Procurement asks a different question: Can you show me how it is controlled?

The draft proposes eight principles covering Human Oversight, Accountability and Redress, Data Governance, Technical Safety and Security, Fairness and Transparency, Sustainability, Societal Benefit, and Training and Literacy.

At project level, it proposes human oversight teams, defined intervention points and named accountable officers. It also calls for an internal inventory of AI use as a single source of truth, alongside a public register of AI systems.

That starts to turn broad ideas about “responsible AI” into things someone may actually have to document.

  1. Who decided what the system is allowed to do?

  2. Where does its authority stop?

  3. Who checks the output?

  4. What happens when it gets something wrong?

  5. What evidence is retained?

  6. Who is accountable for the result?

There are also parts of the draft worth challenging.

It states that Data Protection Impact Assessments should be mandatory before any AI project. That goes further than Article 35 of UK GDPR, which requires a DPIA where processing is likely to result in a high risk to people’s rights and freedoms.

It also says the EU AI Act is “not concerned with businesses who use AI for efficiencies”. That is too broad. Whether the Act applies depends on the organisation’s role, the system, its use and territorial scope. Business use is not automatically outside it.

Those points are precisely why consultation matters. The wider direction, though, is difficult to miss.

AI governance is moving from principle towards evidence.

For businesses supplying AI into public services, the question may increasingly be less “Do you use AI responsibly?” and more: “Show us.”

This is where a contAIn™ R.E.S.P.E.C.T. Data Check fits. It gives businesses a practical way to examine the responsibilities around data, oversight and accountability before somebody else, a buyer, regulator or client, asks them to evidence it.

Northern Ireland’s consultation closes at 5.00 pm on Wednesday 7 October 2026.

consultations.nidirect.gov.uk/teo/artificial-intelligence-public-consultation