Responsible AI principles
Technology is selected for a defined business problem. Important outputs are reviewed, limitations are documented and people remain accountable for consequential decisions.
Trust Centre
Practical controls for human-supervised AI and digital systems
Trust is designed into discovery, tool selection, implementation and ongoing ownership. Controls are reviewed against the client’s actual requirements; compliance is never assumed.
Discuss your requirementsHuman accountability
Each workflow is designed with a named owner, clear approval points and an escalation route. AI does not become the decision owner.
Data minimisation
Discovery identifies the minimum information needed for the intended outcome. Unnecessary personal or confidential data should not enter a workflow.
Tool-selection standards
Tools are assessed for practical fit, access controls, data handling, cost, maintainability and the client’s existing environment.
Access control
FDS recommends least-privilege access, role-appropriate permissions and removal of access when it is no longer required, subject to the selected platform and client configuration.
Retention considerations
Retention needs are reviewed during discovery and should match the purpose, client policy and selected provider. FDS does not assume one period fits every organisation.
Third-party transparency
The providers and integrations needed for an agreed solution are documented. Their terms, hosting and data controls remain subject to the selected service and client configuration.
Security approach
Implementations are designed to support secure defaults, server-side secret handling, input validation, controlled logging and appropriate separation between public demonstrations and real client workflows.
Incident escalation
Operational owners and escalation routes are recommended according to client requirements. Technical or data incidents should be contained, recorded and reviewed by the appropriate responsible person.
Accessibility commitment
FDS aims to design digital experiences around WCAG 2.2 AA practices, including keyboard access, visible focus, semantic structure, clear errors and reduced-motion support. This is not a claim of formal certification.
Automation governance
Automation rules, exceptions, approval points, monitoring and failure handling are documented before consequential actions are enabled.
Limits of AI-generated outputs
AI output may be incomplete, incorrect or inappropriate. It requires review against approved sources, context and professional judgement before use.
What this Trust Centre does not claim
FDS does not claim automatic GDPR, ISO, SOC 2 or sector-specific compliance. Appropriate legal, regulatory, contractual and technical requirements are reviewed during discovery and remain subject to the selected platforms, client configuration and specialist advice where required.