Artificial Intelligence – Updates for the Public Sector

The UK Government has recognised the key role that AI will play across the public sector by announcing a £110 million fund to speed up the deployment of AI in the public sector. The use of AI in the public sector brings with it risks, particularly in respect of cyber security and equality law. The UK Government is already implementing AI solutions across government departments (such as the Department for Work and Pensions), which has raised some concerns about the potential for discrimination and bias. Recent scandals in the Netherlands and Spain regarding false accusations of benefit fraud stemming from the implementation of AI detection systems clearly demonstrates such risks.

This blog considers a number of recent AI policy developments relevant to the use of AI in the public sector in the UK, including new guidance and toolkits from the UK Government and the Scottish Government's new AI register.

Increasing AI transparency

The UK Government has recently published its update on AI and Public Standards which explains the progress that the Government has made since publishing its report on Artificial Intelligence and Public Standards in 2020. The update expands on the principles that will govern the use of AI in the public sector, and provides an update on the how the Government intends to support the public sector in implementing AI products.

The key updates from the report are:

1. The establishment of a steering committee compromising of Government regulators and representatives to support knowledge exchange and the coordination of AI governance.
2. The deadline has now passed for regulators to respond to the Government with guidance on how they intend to implement the five principles to regulate AI (being Safety, security and robustness; Appropriate transparency and explainability; Fairness; Accountability and governance; and Contestability and Redress), which were proposed by the AI White Paper (and which are further discussed in our previous blog, here).
3. Consultations are ongoing for the establishment of a multidisciplinary risk monitoring and assessment team which will monitor the changing risk landscape of AI.
4. The Human Rights Commission has published guidance on the use of AI in public services, which includes a checklist to assist public bodies in considering how they can use AI while complying with the Public Sector Equality Duty.
5. The launch of the Algorithmic Transparency Recording Standard (ATRS) for all central government departments. The ATRS helps public sector organisations provide clear information about the AI tools they use and why they use them. The Government intends for this reporting obligation to be rolled out to the broader public sector in future.
6. The launch of the AI Management Essentials scheme which sets a minimum best practice standard for companies selling AI products and services. The Government also intends this to be a mandatory requirement of public sector procurement in the future to achieve responsible innovation.
7. The Crown Commercial Service has launched a purchasing system for AI, which establishes a baseline ethical standard that suppliers must meet to join the pool of pre-qualified suppliers.
8. The Government has recommended that public sector bodies establish governance arrangements to manage ethical risks and ensure regulatory compliance.

Scottish AI Alliance

The Scottish Government has announced that Scotland will become the first part of the UK to make it mandatory to record any use of AI within the public sector. The new Scottish AI Register – the latest initiative of the Scottish AI Alliance - is a publicly accessible database which will contain information about the use of AI technology in projects developed by public bodies in Scotland. The register will include detailed information of what processes are being used, and how data is helping public sector bodies make better decisions and improve public services.

The Register is currently a voluntary scheme, but the Scottish Government intends on making it mandatory across the public sector in Scotland, starting with Scottish Government departments. The date by which all public sector bodies will have to report on their AI use has not yet been announced.

Responsible AI Toolkit

The UK Department for Science, Information and Technology (DSIT) has published the responsible AI toolkit which is a collection of guidance to support organisations and practitioners to safely and responsibly develop and deploy AI systems in accordance with certain "AI assurance principles".

The AI assurance principles encourage organisations to constantly evaluate and measure the performance and functionality of their AI systems and are suggested as forming part of the broader umbrella of ESG goals that an organisation may have. The toolkit recognises a variety of methods that can be used to ensure the safe development of AI, including conformity assessments, measurements against the UK GDPR, and impact assessments. The AI toolkit suggests a holistic approach to evaluating AI and warns against adopting a standardised "tick box" exercise.

The updated AI toolkit includes guidance on responsible AI use in recruitment and the ATRS template for recording the use of AI. DSIT have announced that this toolkit will be updated over time to reflect the changing regulatory landscape.

Summary

It is clear from the latest published updates, initiatives and guidance that both the UK and Scottish Governments recognise the concerns which organisations, particularly those in the public sector, may have in relation to the safe use and development of AI systems and to ensure that the risks of AI are adequately managed. Such risks are particularly applicable to public sector organisations, which are under heightened scrutiny, many of which have access to large volumes of sensitive information and data, which AI systems may in the future gain access to. Effective evaluation procedures will need to be built into the heart of public sector operations and from the moment at which an organisation decides to implement AI technology into a project.

Public bodies should, if they haven't already, start to prepare AI evaluation and recording procedures and mechanisms, and policies and specific training should be put in place to ensure that staff and employees working within public sector organisations understand the technology which they are being asked to use and are aware of the risks. As both Governments intend to make various AI schemes mandatory for public sector bodies in due course (e.g., the AI management essentials scheme, ATRS, and the AI Register in Scotland), public bodies will benefit from being prepared for these changes.

Should you wish to discuss any of the above, or legal issues pertaining to AI systems generally, in more detail - please contact Martin Sloan, Ally Burr, or your usual Brodies contact.