A national AI skills benchmark is coming: what school leaders should do now

The Department for Education is moving towards something schools have so far largely had to invent for themselves: a common benchmark for the digital, data and AI skills expected of staff.

A £300,000 contract has been awarded to Educate Ventures Research to develop a new “digital and data skills pathway” for schools in England. The work is intended to define both minimum and aspirational standards for the skills staff need to use technology, data and artificial intelligence safely and effectively.

According to Tes, the pathway is expected to cover classroom practice and school management, including the use of technology to improve pupil engagement and outcomes, promote inclusion and reduce workload. It should also give schools a clearer basis for judging the quality of professional development and AI training offered to staff.

The contract is expected to run from September 2026 to the end of March 2027.

For school leaders, the important question is not simply what the eventual framework will contain. It is what to do before it arrives.

Don’t over-engineer your own framework yet

Many schools are currently trying to move beyond basic AI policies towards a more structured model of staff development. That is understandable. Leaders want to know what teachers, support staff and managers should be able to do, how expectations should differ by role and experience, and what good progression in AI confidence might look like.

Until now, there has been no single national reference point for that work.

That is about to change.

This makes the next six months a poor time to invest heavily in a highly detailed, bespoke whole-school AI competency framework. A school could spend considerable time defining levels, descriptors, assessment tools and CPD pathways, only to discover in 2027 that the national model uses different language, priorities or progression points.

The more sensible approach is to separate what is urgent now from what can reasonably wait.

Schools need clear expectations for safe and responsible AI use today. They do not necessarily need a fully developed competency architecture today.

What schools should establish now

A short set of baseline expectations is still essential. Staff should know, for example, which tools are approved for use, what information must never be entered into public AI systems, when AI-generated material needs checking, and where professional judgement remains non-negotiable.

At minimum, schools should consider making clear that staff are expected to:

• protect personal, confidential and safeguarding information;
• use only tools that meet the school’s data-protection and security requirements;
• verify important factual claims and outputs rather than treating generated content as authoritative;
• retain professional responsibility for decisions, feedback and communications produced with AI assistance;
• understand that apparently fluent output may still be inaccurate, biased or inappropriate;
• follow school expectations around copyright, attribution and the use of pupil work;
• escalate concerns where an AI product behaves in ways that could create safeguarding, privacy or reputational risks.

These are not sophisticated competency levels. They are sensible operational guardrails.

Once the DfE pathway is published, schools can then map more formal staff-development expectations against a national reference point rather than building everything from scratch.

The framework may also change how schools buy CPD

One of the more useful aspects of the planned pathway is that it is intended to help schools judge the quality of professional development.

The AI training market has expanded rapidly, and quality is variable. Some courses are little more than demonstrations of prompting techniques. Others focus heavily on particular commercial tools. What leaders increasingly need is training that develops judgement: when AI is useful, when it is not, what risks need managing and how its use connects to teaching quality, inclusion, workload and organisational decision-making.

A national skills pathway could give schools a stronger basis for asking whether a CPD provider is actually developing recognised capability or simply selling enthusiasm.

Until that benchmark exists, leaders should be cautious about purchasing large, multi-year AI training programmes built around a provider’s proprietary competency model. Shorter, practical and reviewable programmes are easier to adapt when the national framework becomes clearer.

There is already a useful DfE benchmark for products

Schools do not, however, need to wait for national guidance in every area.

The DfE has already published generative-AI product safety standards, updated in January 2026. These are aimed primarily at developers and suppliers, but the Department explicitly says schools and colleges may also find them useful when assessing whether AI products are safe for educational use.

The standards cover areas including stated purpose, educational use cases, filtering, monitoring, security, privacy and data protection, intellectual property, governance and risks to children’s cognitive, emotional and social development.

For leaders evaluating AI products, these standards offer a useful starting checklist now.

Rather than asking only, “What can this tool do?”, procurement conversations should include questions such as: What data does it collect? Where is that data processed? Can administrators control permissions? How are harmful outputs filtered? What monitoring and reporting is available? What evidence supports claims about educational impact? How does the supplier manage updates, security and safeguarding risks?

Those questions are likely to be more important than a polished product demonstration.

A sensible position for 2026–27

The direction of travel is becoming clearer. Government is moving from broad encouragement around AI towards more specific expectations about capability, safety and implementation.

For school leaders, that suggests a pragmatic position for this academic year:

Set basic safe-use expectations now. Use the existing DfE product-safety standards when assessing suppliers. Keep staff development practical and adaptable. Then align your longer-term AI and digital competency framework with the national pathway when it is published.

There is no advantage in standing still while waiting for government guidance. But there is equally little value in spending heavily to create a detailed framework that may soon need to be rebuilt.

The task for leaders is therefore not to predict the final DfE standard. It is to make sure the school is safe, thoughtful and ready to adopt it when it arrives.