Responsible AI use

1. Purpose and scope

This statement sets out principles Syspera & Company Limited considers important when using or advising on artificial intelligence (AI). It is a public-facing set of principles, not a claim that every AI control described here has already been implemented in every project. Specific controls should be defined for each use case and engagement.

2. Human accountability

AI may support analysis, creativity, decision-making and operational work, but responsibility remains with the people and organisations that design, procure, deploy and use it. High-impact or consequential decisions should receive appropriate human review and a clear route for challenge or escalation.

3. Purpose and proportionality

AI should be used for a defined, legitimate purpose where the expected benefits are proportionate to the risks. Organisations should consider whether a simpler or less risky approach would achieve the same objective, and should avoid deploying AI merely because the technology is available.

4. Fairness and inclusion

AI systems can reproduce or amplify bias. Where relevant, systems should be assessed for discriminatory effects, representation gaps, accessibility, and impacts on affected groups. Testing and mitigation should be proportionate to the use case and repeated when data, models or operating conditions change.

5. Privacy and confidentiality

Personal, confidential, client or otherwise restricted information should not be entered into an AI system unless the system is approved for that information, the use is authorised, and applicable privacy, contractual and security requirements are met. Data minimisation, retention limits and appropriate controls should be considered.

6. Security and resilience

AI systems may introduce risks including prompt injection, data leakage, model manipulation, insecure integrations, unauthorised actions and misuse of generated outputs. Relevant systems should be threat-modelled, tested, monitored and protected through controls appropriate to their risk and operating context.

7. Transparency and communication

Where AI materially shapes a service, output or decision, organisations should consider what users and affected people need to know to understand the role and limitations of the system. AI-generated content should not be presented as verified fact without suitable review.

8. Accuracy and intellectual property

Outputs should be checked for factual reliability, context, bias, confidentiality and suitability before they are relied upon or published. Users should consider copyright, licensing, attribution, provenance and contractual restrictions when providing inputs to or using outputs from AI tools.

9. Governance and oversight

Responsible deployment should include clear ownership, approved tools, risk assessment, access management, staff guidance, incident handling, documentation and periodic review. Higher-risk uses may require additional testing, independent assurance or formal approval before deployment.

10. Monitoring and improvement

AI risks evolve as models, data, integrations and use cases change. Organisations should monitor performance and incidents, provide a way to report concerns, and update controls when new risks or failures are identified.

11. Questions

For questions about this statement or AI-related advisory enquiries, contact hello@syspera.com. Project-specific AI requirements should be confirmed in the relevant engagement documentation.