As regulation becomes more complex, the challenge is no longer just drafting rules. It is making sure those rules are accurate, consistent, clear to navigate and capable of delivering the intended policy outcome.
In my recent Digital Leaders Webinar Authoring Official Information, I explored how structured data and AI tools can help improve the way official information is authored, managed and delivered. The focus was not on AI as a replacement for expertise, but on how it can support better, faster and more reliable regulatory drafting.
Why this matters
Regulators are under pressure to do several things at once. They need to support growth, remove unnecessary barriers, guide behaviour effectively and make sure regulation is accessible and understandable for the organisations expected to follow it.
That is not easy in a landscape where drafting is still often highly manual, processes can be lengthy, and unintended consequences may only become visible late in the day.
When issues are found after publication, the cost is high. Changes may need to be revisited, guidance adjusted, and confidence rebuilt. Getting things right first time matters.
The problem with traditional drafting processes
In my experience, regulatory drafting often involves complex rulebooks, overlapping provisions, linked guidance and multiple rounds of review. It can be difficult to see clearly how one change affects another, or how a new policy intention interacts with the wider framework.
This creates several familiar challenges:
-long and resource-intensive drafting cycles
-risk of inconsistency across rules and guidance
-difficulty identifying conflicts, dependencies and unintended consequences
-limited visibility of how changes affect different stakeholders
The result is a process that works, but not always as efficiently or as confidently as it could.
Why structured content is the foundation
One of the key points I made during the session is that AI only becomes truly useful when content is properly structured.
Rather than thinking about regulation as static documents, there is a growing opportunity to treat it as structured, enriched data. That means content is not only readable by people, but also usable by machines. With metadata, consistent content types and a single source of truth, organisations can start to apply AI much more effectively.
This shift matters because it allows you to:
-search and analyse content more intelligently
-understand links between rules, guidance and definitions
-manage updates more consistently
-publish to multiple channels from the same source
-apply AI tools with greater accuracy and control
Without this foundation, AI has much less to work with.
Where AI can support the drafting lifecycle
In the session, I focused on three main areas where AI can add value.
1. Research and summarisation
Before drafting begins, significant time is often spent understanding policy intent, reviewing current provisions and identifying what may need to change.
AI can support this stage by analysing policy instructions, mapping key concepts, identifying relevant provisions and surfacing possible conflicts or dependencies. This helps create a drafting pack that gives legal and policy teams a much clearer starting point.
2. Authoring and editing
Within a structured content management environment, drafting and editing can become more controlled and more transparent.
Changes can be made directly in the relevant sections, workflows can support approvals and audit trails, and content can be enriched consistently as part of the process. This improves governance and reduces the risk of fragmentation across documents and teams.
3. Post-drafting checks
This is one of the most valuable areas for AI support.
Once drafting has been completed, AI can help check whether the changes do what they were intended to do. That includes identifying:
-affected obligations and definitions
-cross-reference issues
-inconsistencies with existing provisions
-possible legal or operational risks
-impacts on different stakeholders
-areas that may need further review before publication
Used well, these tools help teams catch problems earlier and move into governance and approval with greater confidence.
Supporting better policy outcomes
Ultimately, the goal is not to introduce technology for its own sake. It is to support better policy outcomes.
That means making it easier to:
-remove barriers where appropriate
-improve clarity and accessibility
-reduce the risk of unintended consequences
-deliver regulation that is easier to find, use and understand
Over time, this creates what I would describe as a virtuous cycle where each iteration of policy builds on a more reliable and better-structured foundation.
Trust still matters
A strong theme throughout my session was that trust remains essential.
AI must be transparent, explainable and grounded in secure, well-managed data. It should support expert judgement, not replace it. At TSO, we align this approach with the UK Government AI Playbook principles, with a clear focus on ethical use, human oversight and auditability.
That is especially important in high-trust environments like regulation, where accuracy and accountability are critical.
Final thought
The opportunity here is significant.
By combining structured content, strong workflows and carefully applied AI, regulators can improve both efficiency and confidence in the drafting process. The result is not just faster work, but better-quality outcomes.
For organisations responsible for authoring official information, that is a meaningful step forward.
Blog post written by Alan Blanchard, TSO Business Development Director.
