Common questions about AI adoption

Browse common questions about adopting AI in government. Each answer includes a snippet from the AI Playbook for the UK Government and a deep link to the relevant section.

This explorer currently covers one source. More public sector AI guidance will be added over time.

Getting started

What is AI, and what are its main limitations in a government context?

You should learn what AI can and cannot do before using it. AI systems currently lack reasoning and contextual awareness, are not guaranteed to be accurate, and have limitations that vary by tool and context.

From the guidance

AI systems currently lack reasoning and contextual awareness and their limitations vary depending on the tools you use and the context in which they operate. AI systems are also not guaranteed to be accurate.

Source: AI Playbook for the UK Government
Section: Principle 1: You know what AI is and what its limitations are

Read this in the AI Playbook (opens in new tab)

When is AI the right tool for the job, and when should I use something else?

Choose the most appropriate technology for the need. Be open to AI where it helps, but also open to concluding that established technologies are a better fit.

From the guidance

However, you should also be open to the conclusion that, sometimes, AI is not the best solution for your problem: it may be more easily solved with more established technologies.

Source: AI Playbook for the UK Government
Section: Principle 6: You use the right tool for the job

Read this in the AI Playbook (opens in new tab)

What kinds of AI use cases work well in government?

Use cases should be led by business and user needs. Focus on problems that only AI can solve well, or where AI offers significant advantages — for example pattern detection in large datasets, complex dynamic decisions, or personalisation.

From the guidance

Once you’ve identified the challenges and opportunities through user research, focus on use cases that can only be solved by AI or where AI offers significant advantages over existing techniques.

Source: AI Playbook for the UK Government
Section: Identifying use cases for AI

Read this in the AI Playbook (opens in new tab)

Which AI use cases should I avoid?

Avoid fully automated decision making for significant decisions, and do not use AI on its own in high-risk areas that could harm health, safety, fundamental rights or the environment.

From the guidance

These include but are not limited to: fully automated decision making: be cautious about any use case involving significant decisions, such as those involving someone’s health or safety. high-risk or high-impact applications: AI should not be used on its own in high-risk areas which could cause harm to someone’s health, safety, fundamental rights or the environment

Source: AI Playbook for the UK Government
Section: Use cases to avoid

Read this in the AI Playbook (opens in new tab)

Do I need an AI strategy or governance board before I start my first project?

Support structures do not need to be fully mature before your first project, but you should have enough control to use AI safely. Establish strategy, principles, governance, communication, sourcing and training as you go.

From the guidance

These structures do not need to be fully mature before your first project. Your experience in your first AI project will shape the way you organise these structures.

Source: AI Playbook for the UK Government
Section: Creating the AI support structure

Read this in the AI Playbook (opens in new tab)

What skills and roles do I need on an AI project team?

Build a multidisciplinary team covering user needs, delivery, data, engineering, design, and legal/commercial/security/ethics expertise. Balance technical and domain expertise and include diverse viewpoints to help spot bias.

From the guidance

Your team should include or be able to collaborate with: business leaders and experts who understand the context and impact on users and services; data scientists who understand the relevant data, how to use it effectively, and how to build and train and test models; software engineers who can build and integrate solutions; user researchers and designers who can help understand user needs and design compelling experiences; legal, commercial and security colleagues, as well as ethics and data privacy experts, who can help you make your AI solution safe and responsible

Source: AI Playbook for the UK Government
Section: Building the team

Read this in the AI Playbook (opens in new tab)

Where can civil servants get training on AI?

Free AI courses are available on Civil Service Learning and through Government Campus learning frameworks. Senior civil servants can also use the Digital Excellence Programme AI course.

From the guidance

AI courses are freely available within Civil Service Learning, as well as a series of AI courses from the Government Campus which can be accessed through the learning frameworks.

Source: AI Playbook for the UK Government
Section: Learning resources

Read this in the AI Playbook (opens in new tab)

Lawful, ethical and responsible use

Can I use AI to process personal data?

Yes, but only lawfully and with data protection advice from the start. AI systems can process personal data, so you must protect it, comply with data protection law, and minimise privacy intrusion from the outset.

From the guidance

You should seek data protection advice on your use of AI. This may be from your lawyers or your data protection officer. AI systems can process personal data, so you need to consider how you protect this personal data, be compliant with data protection legislation, and minimise the risk of privacy intrusion from the outset.

Source: AI Playbook for the UK Government
Section: Principle 2: You use AI lawfully, ethically and responsibly

Read this in the AI Playbook (opens in new tab)

Do I need a Data Protection Impact Assessment (DPIA) before using AI?

Before implementing AI solutions you need to undertake a DPIA. UK GDPR also requires a DPIA for certain high-risk processing, and the ICO requires one when using innovative technologies.

From the guidance

Before implementing AI solutions, you need to undertake a data protection impact assessment (DPIA). This involves an assessment of data protection and privacy risks, and the implementation of appropriate technical and organisational measures to sufficiently mitigate them.

Source: AI Playbook for the UK Government
Section: Lawfulness and purpose limitation

Read this in the AI Playbook (opens in new tab)

Can I reuse existing personal data to train or run an AI system?

Only if the new purpose is compatible with the original purpose for collection. Assess expectations, data type, impact on people, and whether extra safeguards are needed.

From the guidance

AI systems often reuse personal data for new purposes that are different from those for which it was originally collected. This may cause tension with the purpose limitation of the UK GDPR. Repurposing of personal data is only legitimate if a new purpose is ‘compatible’ with the purpose for which the data was originally collected.

Source: AI Playbook for the UK Government
Section: Lawfulness and purpose limitation

Read this in the AI Playbook (opens in new tab)

Can AI make automated decisions that affect people?

Solely automated decisions with legal or similarly significant effects are restricted under UK GDPR Article 22. Where AI affects someone’s legal status or rights, it must only support decisions made by a human.

From the guidance

Article 22 currently prohibits decision(s) based solely on automated processing that have legal or similarly significant consequences for individuals. Services using AI that affect a person’s legal status or their legal rights must only use AI to support decisions that must be made by a human decision maker.

Source: AI Playbook for the UK Government
Section: Human oversight

Read this in the AI Playbook (opens in new tab)

How much human oversight do I need when using AI in decision making?

You need meaningful human control at the right stages, including validation of high-risk decisions and ways for users to report issues and trigger human review.

From the guidance

You need to monitor the AI’s behaviour and have plans in place to prevent any harmful effects on users. This includes ensuring that humans validate any high-risk decisions influenced by AI and that you have strategies for meaningful intervention.

Source: AI Playbook for the UK Government
Section: Principle 4: You have meaningful human control at the right stages

Read this in the AI Playbook (opens in new tab)

Do I need to tell the public when we are using AI or algorithms?

Yes. Be open about where and how algorithms and AI are used in official duties, and clearly identify automated responses such as chatbot replies.

From the guidance

You should be open with the public about where and how algorithms and AI systems are being used in official duties. If you’re a central government department or an arm’s length body within scope, you’re required to use the Algorithmic Transparency Recording Standard (ATRS).

Source: AI Playbook for the UK Government
Section: Principle 7: You are open and collaborative

Read this in the AI Playbook (opens in new tab)

Do I need to use the Algorithmic Transparency Recording Standard (ATRS)?

Central government departments and in-scope arm’s length bodies must use ATRS for algorithmic tools in decision-making. Other public bodies are encouraged to use it too.

From the guidance

If you’re a central government department or an arm’s length body within scope, you’re required to use the Algorithmic Transparency Recording Standard (ATRS). This means you must document information about any algorithmic tools you use in decision-making processes and make it clearly accessible to the public.

Source: AI Playbook for the UK Government
Section: Principle 7: You are open and collaborative

Read this in the AI Playbook (opens in new tab)

How do I manage bias and fairness in an AI system?

AI can reproduce bias from training data and produce unfair outputs. Consider all potential sources of bias across the life cycle, including unrepresentative datasets and unfair deployment impacts.

From the guidance

AI models are trained on data which may include biased or harmful materials. As a result, AI systems may display biases and produce harmful outputs, such as unfair, prejudicial or derogatory representations of groups or individuals. You should consider all potential sources of bias throughout the development life cycle, including unrepresentative data sets and deployment scenarios that have unfair or undesirable impacts.

Source: AI Playbook for the UK Government
Section: Principle 2: You use AI lawfully, ethically and responsibly

Read this in the AI Playbook (opens in new tab)

What equality and human rights issues should I consider when using AI?

Consider Equality Act and Public Sector Equality Duty obligations, and whether AI may affect Convention rights such as privacy or freedom of expression. Assess equality impacts early.

From the guidance

Lawyers can help you navigate the equality issues raised by the use of AI in government – for example, obligations arising under the Equality Act 2010 and the Public Sector Equality Duty. Conducting an assessment of the equality impacts of your use of AI can also be one way to guard against bias, which is particularly important in the context of AI.

Source: AI Playbook for the UK Government
Section: Equality issues

Read this in the AI Playbook (opens in new tab)

Do I need legal advice before starting an AI project?

Yes — seek legal advice early on development and use of AI, including equalities, fairness, intellectual property and other legal issues. Explain aims, capabilities and risks when you contact lawyers.

From the guidance

Your use of AI tools must be lawful and responsible. You should seek legal advice on the development and use of AI and engage with compliance, legal and data protection experts in your organisation early in your journey, including during product development.

Source: AI Playbook for the UK Government
Section: Principle 2: You use AI lawfully, ethically and responsibly

Read this in the AI Playbook (opens in new tab)

Security and safe use of tools

Can I put official or unpublished information into public tools like ChatGPT?

No. When using public AI applications you must not enter official information unless it has been published or is cleared for publication.

From the guidance

When using public AI applications, you must not enter official information unless it has been published or is cleared for publication.

Source: AI Playbook for the UK Government
Section: Public AI applications and web services

Read this in the AI Playbook (opens in new tab)

Can I use Microsoft Copilot, Slack GPT, or similar embedded AI features at work?

Only after understanding the product architecture and vendor mitigations, and after speaking with your security team. Embedded AI features bring their own security concerns.

From the guidance

Before adopting any of these products it’s important to understand the underlying architecture of the solution, and what mitigations the vendor has put in place for the inherent risks associated with AI.

Source: AI Playbook for the UK Government
Section: Embedded AI applications

Read this in the AI Playbook (opens in new tab)

Are AI meeting transcription tools allowed?

Treat them as a serious data-leakage risk. Meeting organisers should verify attendees and state that third-party transcription tools are not allowed.

From the guidance

There has also been a proliferation of AI transcription tools that are capable of joining virtual meetings and transcribing meeting notes. These present a serious risk of data leakage as they silently upload meeting recordings to an AI service for transcription and analysis. When hosting virtual meetings, organisers should verify the identity of all attendees and state up front that the use of third-party meeting transcription tools is not allowed.

Source: AI Playbook for the UK Government
Section: Embedded AI applications

Read this in the AI Playbook (opens in new tab)

Should I use a public AI API, a privately hosted model, or a managed platform?

It depends on control and risk. Public APIs still send data to a provider; private hosting keeps data in your environment but you own security and ops; managed platforms can offer private instances with stronger retention controls.

From the guidance

By running a model in your own private cloud infrastructure, you ensure that data never leaves an environment that you own.

Source: AI Playbook for the UK Government
Section: Privately hosted AI models

Read this in the AI Playbook (opens in new tab)

What are the main security risks of using AI in government?

AI-specific risks include data/model poisoning, data leakage, insecure AI tooling, prompt injection, perturbation attacks and hallucinations, as well as amplification of existing cyber risks.

From the guidance

Some threats – such as data poisoning, perturbation attacks, prompt injections and hallucinations – are specific to AI. However, AI systems can also amplify generic risks such as phishing and cyber attacks.

Source: AI Playbook for the UK Government
Section: Principle 3: You know how to use AI securely

Read this in the AI Playbook (opens in new tab)

How do I stop an AI system from leaking personal or sensitive data?

Control what data the model can access, prefer approaches that preserve user access controls (for example RAG/in-context learning), and apply additional security controls when using organisational data.

From the guidance

If you use your own data with an AI model, you immediately increase the data security risk and you need to apply additional security controls to stop data leakage and privacy violations.

Source: AI Playbook for the UK Government
Section: Working with your organisational data

Read this in the AI Playbook (opens in new tab)

How should I handle prompt injection and other generative AI-specific threats?

Assume prompts can subvert system instructions. Use filtering, logging and audit, and keep a human in the loop before automated actions are carried out.

From the guidance

Fundamentally, a generative AI model cannot distinguish between the user prompt and these system instructions because both are just seen as input to the model. A hacker can exploit this flaw by crafting special prompts that circumvent the system instructions, causing the model to respond in an unintended way.

Source: AI Playbook for the UK Government
Section: Prompt injection

Read this in the AI Playbook (opens in new tab)

Can I trust generative AI outputs, or do they hallucinate?

Do not trust generative AI to produce factual content uncritically. Models can generate plausible but false information; train users not to rely exclusively on these outputs.

From the guidance

Fundamentally, generative AI models cannot be trusted to produce factual content. Any generative AI services that output generated content directly to the public – for example, an LLM-powered chatbot giving advice on a government website – would be prone to hallucination and could lead to someone being misled about a government service, policy or point of law.

Source: AI Playbook for the UK Government
Section: Hallucinations

Read this in the AI Playbook (opens in new tab)

Buying and building

Should I buy an AI product or build one in-house?

Decide based on the problem and commercial advice. Options include off-the-shelf products, AI bolted onto existing technology, outsourced builds, or co-creating with suppliers.

From the guidance

This might be an off-the-shelf product, an existing technology with bolt-on AI elements (paid or free), outsourcing AI builds (if applicable), or co-creating AI with suppliers.

Source: AI Playbook for the UK Government
Section: Specifying your requirements

Read this in the AI Playbook (opens in new tab)

When should I involve commercial colleagues in an AI project?

From the start. Get commercial advice early on partners, pricing, products and services, and keep ethical expectations consistent for in-house and procured systems.

From the guidance

AI is a rapidly developing market, and you should get specific advice from commercial colleagues on the implications for your project. Reach out to them early in your journey to understand how to use AI in line with commercial requirements.

Source: AI Playbook for the UK Government
Section: Principle 8: You work with commercial colleagues from the start

Read this in the AI Playbook (opens in new tab)

How do I write a business case for an AI project?

Use the Green Book for larger investments (typically approaching £10m) and consider GDS agile business case guidance for smaller ones. Engage stakeholders first on whether AI is needed.

From the guidance

Consider following the Treasury’s Green Book (2022) guidance to create a fully fledged, five-part business case. This may be a requirement depending on the scale of your project and investment.

Source: AI Playbook for the UK Government
Section: AI business cases

Read this in the AI Playbook (opens in new tab)

What spend controls or approvals apply to AI and digital projects?

Digital and technology spend above £100,000 for public-facing services and £1 million otherwise must be assured through your assurance boards. Follow GDS spend approval guidance.

From the guidance

When working on your business case, note that it’s mandatory to assure all digital and technology spend above £100,000 for anything public facing and £1 million for anything else, through your assurance boards.

Source: AI Playbook for the UK Government
Section: AI business cases

Read this in the AI Playbook (opens in new tab)

How do I specify requirements when buying AI?

Start from the problem statement, cover data strategy/quality/bias, demand transparency about the supplier’s AI approach, plan for maintenance, IP, liabilities and avoiding vendor lock-in.

From the guidance

When drafting requirements for AI, you should: start with your problem statement; highlight your data strategy and requirements; focus on data quality, bias (mitigation) and limitations; underline the need for you to understand the supplier’s AI approach; consider strategies to avoid vendor lock-in

Source: AI Playbook for the UK Government
Section: Specifying your requirements

Read this in the AI Playbook (opens in new tab)

Who owns the intellectual property if we develop or procure an AI solution?

Decide ownership and ongoing use rights at the outset, including how infringement risk and liability are shared between parties.

From the guidance

For example, you should consider at the outset: which parties will own which parts of any intellectual property generated during the project; which parties will have ongoing rights to use any intellectual property that is generated (and on what basis); how the balance of risk and liability should be determined between the parties, as this will be relevant to any claims for infringement of third party intellectual property

Source: AI Playbook for the UK Government
Section: Intellectual property, including copyright

Read this in the AI Playbook (opens in new tab)

How do I avoid vendor lock-in when buying AI?

Build exit and portability into requirements and contracts from the start, including understanding the supplier’s approach and planning for transfer to successor suppliers.

From the guidance

consider strategies to avoid vendor lock-in

Source: AI Playbook for the UK Government
Section: Specifying your requirements

Read this in the AI Playbook (opens in new tab)

Delivery, assurance and operations

Do AI projects still need to meet the government Service Standard?

Yes. If you develop a service, you must use the government Service Standard, alongside wider technology and cloud security guidance.

From the guidance

If you develop a service, you must use the government Service Standard.

Source: AI Playbook for the UK Government
Section: Principle 5: You understand how to manage the full AI life cycle

Read this in the AI Playbook (opens in new tab)

How should I do user research for an AI product?

Use user research to check AI is the right tool, define metrics, prepare/evaluate data and outputs, assess usability and trust, and monitor the live service — keeping humans in the loop.

From the guidance

Doing UR for an AI project helps you keep the human in the loop and understand the human intelligence that the AI will replicate or imitate.

Source: AI Playbook for the UK Government
Section: User research for AI

Read this in the AI Playbook (opens in new tab)

How do I monitor an AI system once it is live?

Put ongoing performance monitoring in place, evidence that the system is operating as expected, and manage model updates through a controlled release process that can be rolled back.

From the guidance

Once you’ve released your AI system for use and it’s operational, you should have ongoing performance monitoring in place. This will ensure your system is operating as expected, and you should be able to provide evidence of this.

Source: AI Playbook for the UK Government
Section: Operational monitoring

Read this in the AI Playbook (opens in new tab)

What happens if the model drifts or starts performing worse over time?

Monitor for drift. Environments change over time and may require retraining or a new model; catch this early to reduce disruption.

From the guidance

As systems and environments evolve, the current process may diverge sufficiently from the training period of the AI system. This is known as model drift and may require retraining or implementation of a new model within the AI system. Close monitoring is essential so that you can catch this as early as possible and reduce possible disruption to the AI system.

Source: AI Playbook for the UK Government
Section: Operational monitoring

Read this in the AI Playbook (opens in new tab)

How do I keep an inventory of AI systems in my organisation?

Maintain a live AI/ML systems inventory covering purpose, risks, data, ownership and key dates, in addition to ATRS transparency records where required.

From the guidance

To provide a comprehensive view of all deployed AI systems within an organisation or programme, organisations should set up an AI and machine learning (ML) systems inventory. This is in addition to the Algorithmic Transparency Recording Standard (ATRS) that all government departments and certain arm’s length bodies must use

Source: AI Playbook for the UK Government
Section: Creating an AI systems inventory

Read this in the AI Playbook (opens in new tab)

Who is accountable if an AI system causes harm or makes a bad decision?

Your organisation needs clear ownership of risk and responsibility for mitigations and compliance. Connect with assurance teams early and document review and escalation routes.

From the guidance

Accountability is a key principle that establishes ownership of risk, responsibility for mitigations, compliance with legislation, the ability to demonstrate compliance, and high standards for privacy.

Source: AI Playbook for the UK Government
Section: Accountability

Read this in the AI Playbook (opens in new tab)

How do people challenge or seek redress for an AI-influenced decision?

Build contestability and redress into design so people can challenge outcomes and seek remedy. This sits alongside transparency, explainability and meaningful human oversight.

From the guidance

You should have systems in place that allow users to report issues and prompt a human review.

Source: AI Playbook for the UK Government
Section: Principle 4: You have meaningful human control at the right stages

Read this in the AI Playbook (opens in new tab)

Collaboration and reuse

How do I find out what other departments are already doing with AI?

Join cross-government communities such as the AI community of practice, engage departments tackling similar problems, and review ATRS records and published case studies.

From the guidance

You should make use of existing cross-government communities where there is a space to solve problems collaboratively, such as the AI community of practice. You should also engage with other government departments that are trying to address similar issues and reuse ideas, code and infrastructure.

Source: AI Playbook for the UK Government
Section: Principle 7: You are open and collaborative

Read this in the AI Playbook (opens in new tab)

Can I reuse AI code, models, or approaches from elsewhere in government?

Yes — the playbook encourages reusing ideas, code and infrastructure, and sharing inventories/case studies through the AI community of practice.

From the guidance

You should also engage with other government departments that are trying to address similar issues and reuse ideas, code and infrastructure.

Source: AI Playbook for the UK Government
Section: Principle 7: You are open and collaborative

Read this in the AI Playbook (opens in new tab)

Should I engage civil society, academia, or industry on my AI project?

Yes where possible. Engaging wider civil society, academia and industry helps ensure AI delivers public benefit and reflects people’s values and concerns.

From the guidance

Where possible, you should engage with the wider civil society including groups, communities, and non-governmental, academic and public representative organisations that have an interest in your project.

Source: AI Playbook for the UK Government
Section: Principle 7: You are open and collaborative

Read this in the AI Playbook (opens in new tab)