From engagement to delivery: supporting responsible AI adoption in councils

In April, the Local AI team ran a series of AI ethics drop-in sessions with councils across England. The aim was simple: to understand what’s helping and what’s getting in the way of responsible AI adoption in local government.

Since then, we’ve used those insights to shape our next phase of work, alongside our ongoing collaboration with the Local Government Association (LGA).

This blog post shares what we heard, what we’ve learned, and what we’re doing next.

What we set out to do

The goal of Local AI’s responsible AI workstream is to help councils use AI safely, fairly and effectively by embedding ethical principles into local government and giving councils the confidence to adopt and scale AI. As a primary aim, we wanted to understand the challenges councils faced when it comes to responsible AI adoption.

So, we set up 4 drop-in sessions to:

• hear directly from councils about their experiences

• test early ideas for our ethics workstream

• start co-designing practical support for the sector

What we heard from councils

We are grateful councils took the time to engage through the drop-in sessions. We heard from 60 councils, which gave us rich insights into the challenges within the sector. There were some clear themes that stood out from all the different sessions:

1. When it comes to ethics, there are no clear minimum expectations that all councils should follow.
2. There is a lot of guidance, which can be confusing, fragmented and hard to navigate.
3. The existing guidance is too high-level and not practical enough to translate into day-to-day decisions.
4. Ethics is usually reduced to just data protection and statutory duties, and there is lack of guidance on the human behavioural impacts and workforce impacts.
5. There is lack of skills and training across councils to procure, use and govern AI tools.

Through our engagement, we learnt that councils are at very different stages of AI adoption. Some are already being innovative with AI, experimenting with tools and governance approaches, while others are still exploring what AI means for their organisation. That means support needs to work across different levels of maturity.

Across all sessions we ran, the same needs came up: clearer practical guidance, usable tools (not just frameworks) and opportunities to learn from peers. Importantly, people weren’t asking for more theory – they were asking for help with implementation.

The drop-in sessions were designed as a co-design exercise and participants shared feedback on some of our early workstream ideas, potential deliverables and gaps in existing support. This has directly influenced what we’re focusing on next.

What we have done since then

Using what we’ve learned, we worked through a shortlisting process to prioritise a set of potential projects and build an evidence base for each. While we’re still refining the detail, the work will be centred around 3 workstreams and will explore the following ideas:

1. Moving from principles to practice – how to clearly define and develop products that set a minimum standard on responsible AI adoption
2. Knowledge sharing and peer support – how to map existing resources instead of creating new ones
3. Training and capability building – how to provide training to councillors, training alongside the roll-out of Local Transcribe and potentially more joined up pilots

What we heard in April reinforced something important: this isn’t a policy problem alone – it’s a practical delivery problem. We’re keen to co-design what we produce and ensure it’s a valuable addition to the existing AI landscape.

What we’re doing next and our collaboration with LGA

Alongside the drop-in sessions, we’ve been working closely with the LGA on their Cyber, Digital, AI and Technology programme for 2026 to 2027.

The aim is to make sure our work is aligned with sector priorities, and councils receive a more coherent offer of support.

We’re therefore excited to announce that we’re providing grant funding to the LGA to co-deliver resources, practical guidance and tools. Some priority projects are:

1. A new AI playbook for local government – the goal is to help councils self-assess their AI adoption maturity, learn from peer examples, and align with national standards (such as the AI Playbook for the UK Government and Data and AI Ethics Framework) while addressing local needs. 
2. Standard assessment templates – the aim is to consolidate data protection, equality impact and other relevant assessment requirements into a standardised template that councils can adopt and use when assessing any AI tool.
3. Expand the AI network into an AI community of practice – the plan is to expand the existing successful Local Government AI network to have more opportunity for collaborations between councils.

Additionally, the Local AI team is working hard on Local Transcribe – an AI transcription and summarisation tool designed to reduce the administrative burden on frontline staff. We are continuing to build Local Transcribe with an emphasis on responsible usage – please watch out for another blog update on that in the coming weeks.

Get involved

The drop-in sessions were just the starting point. We’ll be running co-design workshops to develop and test new deliverables and keeping you updated on our progress through similar blog posts. The next series of co-design workshops running in August and September will be to understand if we can set a baseline on what responsible AI practices are across all councils with the aim to design an AI policy template.

If you’d like to shape future outputs, we’d like to hear from you. Please reach out to us at LocalAI@communities.gov.uk.