Building better
education

DfE and i.AI are working together to enable schools to access safe, high-quality and effective AI tutoring tools. These tools will integrate with lesson content, allowing pupils a new way to explore topics and work through tricky concepts at a level that suits them. The tools will also be useful for teachers, providing rich feedback on their pupil’s understanding of lesson content, and indicating knowledge gaps pupils may have.

We will measure impact through rigorous, real-world trials in schools, including through DfE's Edtech Testbed Programme. We will build evidence on how the tools can be used by teachers and pupils and their impact on learning outcomes for diverse pupils including disadvantaged pupils.

Eight Pioneers have been selected to work with teachers and the government to design safe classroom ready AI tutoring tools, with safety at the forefront. The next generation of AI tools must align to the national curriculum, classroom context, and clearly show how they will support disadvantaged pupils, with the potential to scale and support up to 450,000 pupils in Years 9 and 10.

Pioneer programme work over the coming year will include

1. Co-design with schools. Eight Edtech/AI companies and their partner schools will work to rapidly iterate their products, with an emphasis on exploring how tools can integrate with lesson content, both inside and outside of the classroom.
2. Piloting new functionalities and requirements. Testing how AI tutoring tools can ensure effective curriculum integration and frameworks for sharing privacy-preserving insights.
3. Industry and education sector engagement. Sharing insights between industry and government, sharing emerging challenges and opportunities for the development of integrated generative AI tutoring tools.

Programme Reading Room

The Department for Education will be hosting the AI Tutoring Tools Reading Room which will make key insights from the Pioneers Programme available to the wider market. 

This is to enable other organisations to access the relevant insights generated through the Programme's research and development work, without requiring the wider market to duplicate the research investment undertaken under the Programme. 

The Programme Reading Room will be hosted on the DfE Content Store LINK and further updates will be provided via the content store as materials become available.

Any further updates will be shared on this webpage in the Summer of 2026.

In the context of AI, a benchmark is a structured way of testing how a model performs against a defined set of tasks, risks, and quality standards. Rather than relying on marketing claims or general-purpose performance scores, benchmarks evaluate AI systems using realistic scenarios, clear success criteria, and consistent scoring methods.

Sovereign Education Benchmarks are designed specifically for school and classroom contexts. They test how AI models respond to typical educational interactions, including instructional support and safeguarding-relevant situations. This ensures that performance is measured not only in terms of technical capability of an AI tool, but also pedagogical quality and safety.

Our benchmarks aim to drive progress on pedagogical quality and safety of AI tools. They will be designed to drive up quality across the whole market and rapidly assess new models. The benchmarks will ensure that the technology powering AI tools used in schools meets minimum standards before being trialled. At this stage, they are not intended to offer insights to schools on suitability of tools for use in education.

We are working with teachers to create example classroom interactions and clear scoring criteria for responses. This hands-on input will help ensure the benchmarks are accurate, aligned to teaching standards, and safe to use.

DfE is building a machine-readable digital national curriculum that fully digitises subject content, helping teachers make links within and across subjects, providing a more holistic view of curriculum content. This work will also enable educational tools to integrate directly with curriculum content and create solutions that better fit with what teachers need to teach.

DfE and i.AI are exploring data needs and architecture for input data to give AI tools a rounded understanding of a pupil's journey, and output data where we need to avoid it becoming siloed in a single supplier's tool. This work will integrate with DfE's wider interoperability programme.