Our Use of AI Technology

Background

Onyxsis uses AI only where it has a clear purpose. It can support analysis, drafting, and repeatable work, but it does not replace engineering judgement or our responsibility for the work we deliver.

Purposeful Adoption

We assess each proposed use against the requirement, the data involved, the expected benefit, and the associated risks. Permitted uses include:

• Repeatable Work:
  AI can assist with repeatable tasks, allowing our teams to concentrate on work that requires technical judgement and detailed knowledge of the organisation's environment.
• Supportive Tools:
  AI may assist with data analysis or content drafting. People remain responsible for checking the output against the original requirement and the relevant technical context.

Security & Responsible Use

Our use of AI is subject to the same data-handling, security, and professional responsibilities as the rest of our work:

• Privacy by Design:
  We apply data-minimisation principles and only process information required for the stated purpose, in line with applicable obligations including UK GDPR.
• Bias Risk Assessment:
  Where an AI-assisted output could affect a person or decision, we assess the relevant risks, limitations, and potential for harmful bias.
• Cybersecurity Alignment:
  We apply security controls and supplier assessments appropriate to the data, system, and intended use.

Human Oversight

AI is a tool, not a substitute for human skill and oversight. AI-assisted work is reviewed to a level appropriate to its purpose, risk, and intended use:

• Accuracy & Relevance:
  We check relevant outputs for accuracy, context, and suitability against the original requirement.
• Reviewed Against the Requirement:
  Technical decisions and final deliverables remain the responsibility of the people who understand the organisation's requirement and operating environment.

Transparency & Accountability

We remain accountable for how AI is used in our work:

• Clear Communication:
  Where AI contributes materially to project work, we explain how it is being used and the controls around that use.
• Accountability:
  Our teams remain responsible for AI-assisted outputs and for the decisions made using them.

Review and Governance

AI tools and practices change quickly. We therefore review their use rather than treating an initial approval as permanent:

• Ethical Frameworks:
  We consider relevant codes of practice and guidance when setting controls for safety, fairness, security, and accountability.
• Continuous Improvement:
  We review this policy and our internal controls as technology, guidance, security risks, and project requirements change.

Conclusion

We use AI only where it provides a clear benefit and is suitable for the work involved. Human judgement, appropriate controls, and accountability remain part of the work throughout.