DSIT’s Responsible Technology Adoption (RTA) Unit is pleased to publish its Introduction to AI assurance. This guidance is an accessible introduction that aims to support organisations to better understand how AI assurance techniques can be used to ensure the safe and responsible development and deployment of AI systems.
The introduction supports delivery of the UK’s white paper, A pro-innovation approach to AI regulation (2023) through providing information on tools and processes that can put the white paper’s five cross-cutting regulatory principles into practice. Government’s subsequent AI white paper consultation and response found strong support for these principles and for the use of technical standards and assurance techniques to embed the principles into the AI development process.
Effective AI assurance can support industry to confidently invest in new products and services and to innovate at pace, while managing risk and helping regulators monitor compliance. The development of a robust AI assurance ecosystem also holds economic potential: the UK’s cyber security industry, an example of a mature assurance ecosystem, is worth £4 billion to the UK economy.
The guide covers:
AI assurance in context
This section introduces the background and conceptual underpinnings of AI Assurance to ensure readers understand what it is and how it supports the responsible use of AI.
The AI assurance toolkit
Provides an overview of the core concepts, tools and techniques, stakeholders, and international standards that make-up the assurance ecosystem.
AI assurance in practice
Practical advice on how to embed assurance in the AI development and deployment lifecycle. This includes an overview of the spectrum of AI assurance techniques that can be applied, how assurance can be embedded into the AI lifecycle and how to effectively build governance into organisational processes. This section also provides several practical examples of different assurance techniques.
Key actions for organisations:
A brief overview of practical actions that organisations looking to embed AI assurance can take, these include ensuring internal upskilling and reviewing internal risk management and governance processes.
The guidance will be regularly updated to reflect feedback from stakeholders, the changing regulatory environment, and emerging global best practices. If you’d like more information about AI assurance and how it can be applied to your own organisation, don’t hesitate to contact the AI assurance team at: firstname.lastname@example.org.