Creating law with Artificial Intelligence in the Commonwealth
- DMS
- Jul 12
- 9 min read
The United Arab Emirates HH Sheikh Hamdan Bin Mohammed Bin Rashid Al Maktoum has launched a groundbreaking initiative to use artificial intelligence (AI) in drafting, reviewing, and amending legislation, making it the first country to fully introduce an AI-driven legislative system. A new Regulatory Intelligence Office will aggregate federal and local laws, court rulings, and administrative data into a massive platform to track the real-time impact of laws and even proactively suggest update. Officials claim this AI-powered system could automate up to 70% of lawmaking, dramatically speeding up the legislative cycle while making drafting “faster and more precise.” HH Sheikh Mohammed Bin Rashid Al Maktoum , UAE’s Prime Minister, hailed the system as a “qualitative leap” that will regularly propose amendments to keep laws responsive to societal needs.
While this bold experiment promises efficiency, it raises serious questions for lawmakers and drafters across the world. Can AI draft laws without eroding the democratic processes at the heart of lawmaking?
The Need for Parliamentary Debate and Transparency
The Commonwealth , with its long-standing commitment to the rule of law, should examine how such innovations align with fundamental principles of transparent governance and constitutional democracy. Central to Commonwealth legal culture is the understanding that how laws are made is as important as the laws themselves. As the Commonwealth Legislative Drafting Manual (2017) foreword notes, “the process by which laws are made is determinative of their ultimate quality”, and thus national legislation must be made through transparent procedures. Open parliamentary debate, public scrutiny, and a clear legislative process are not optional niceties – they are safeguards that uphold constitutional freedoms and democratic rights. These processes ensure that legislation is not only well-crafted but also enjoys legitimacy and public trust.
Enthusiasm for AI’s efficiency must not come at the cost of sidelining Parliament or the public. Bypassing democratic legislative processes – rushing bills through on AI’s recommendation without adequate debate or scrutiny – could undermine democracy itself. Parliamentary debate allows elected representatives to interrogate a proposal’s implications for rights and freedoms; transparency lets citizens hold lawmakers accountable. If an AI system’s suggested law changes were adopted with minimal discussion, it would short-circuit these vital checks and balances. The Commonwealth Secretariat’s guidance already emphasises giving legislators ample time and information before laws are passed. For example, standard procedure is that “all Members of Parliament should have had as much opportunity as possible to study the text of a Bill before it is debated in Parliament”, with even more than the minimum notice period required. Such measures exist to preserve deliberation and prevent hasty lawmaking.
In short, no matter how advanced the technology, the role of human deliberation in lawmaking is irreplaceable. The rule of law requires that AI’s input be filtered through the constitutional process – cabinet approvals, readings, committee reviews, and open debates – that confer legitimacy on legislation. AI might assist with drafting, but elected legislators must ultimately decide, justify, and publicly defend the laws, ensuring they meet the standards of rights and justice expected in a democracy.
Risks of Bypassing Democratic Legislative Processes
The UAE’s experiment has prompted admiration and caution from experts. Observers note that while AI can rapidly analyse data and draft text, it lacks the nuanced understanding of human lawmakers. Researchers warn that current AI models suffer from “hallucinations” – they can generate text that is incorrect, incoherent, or contextually inappropriate. Without rigorous human oversight, an AI might propose something “that makes sense to a machine but may absolutely make no sense… in a human society”, as one Oxford researcher cautioned. In legislative terms, this could mean laws that inadvertently conflict with constitutional principles or produce unintended consequences for citizens.
Moreover, AI systems learn from existing data and thus may carry hidden biases. A legislative AI trained on past statutes and cases might reflect outdated or biased norms unless carefully checked. Analysts have pointed out risks like biased training data and the inability to interpret laws as humans do in the UAE’s system. If governments were to lean on AI to draft laws quickly, they might be tempted to cut short the consultative processes that normally catch such problems. Speed should not trump substance: a flawed or rights-infringing bill drafted by AI and rushed into law would be far more costly to democracy than any efficiency gains.
Crucially, executive overreach is always a concern. If an AI regularly “suggests” legal changes, there is a danger that the executive branch (armed with AI tools) could push alterations through faster than the legislature can respond. This scenario could marginalise the legislative branch’s role and weaken the separation of powers. The Commonwealth’s legal framework is built on careful distribution of authority – legislative drafters turn policy into law, but those laws must pass through parliaments to gain force. Any approach that uses AI to skip or compress these steps needs strong guardrails. As AI enters the legal space, experts stress that “integration must be done thoughtfully” with humans “keeping their hands on the steering wheel.” In practice, this means maintaining a human-in-the-loop at every stage: AI might generate a draft, but human legislative counsel and policymakers must review, revise, and approve it, ensuring it aligns with democratic values and legal principles.
Integrating AI Tools into Legislative Drafting Guidelines
The Commonwealth Legislative Drafting Manual (2017) is a comprehensive guide for drafters, covering everything from language usage to the mechanics of bill preparation. It offers a strong foundation of best practices that could be enhanced with AI assistance. Notably, the Manual itself is conceived as a “living document,” inviting “suggestions for corrections, amendments or updates” as drafting evolves. Incorporating AI-aware guidance would be a natural evolution in 2025, aligning the Manual with emerging tools while preserving its core principles.
Plain Language Drafting: The Manual emphasises clarity and simplicity in legislative language. It notes that legal writing should be readily understood by those affected, and that it’s wasteful if laws must be “translated” by lawyers for the public. The general objective of plain English drafting is “to communicate more effectively” by using simple, direct, familiar language, omitting excess verbiage, and avoiding ambiguity. AI can reinforce these goals. For instance, modern NLP tools can analyse a draft bill’s readability and flag complex sentences or jargon. An AI assistant could suggest clearer synonyms for archaic terms or simplify convoluted sentence structures – effectively serving as a real-time plain-language editor. Integrating AI-assisted clarity checks into the drafting process could help drafters meet the Manual’s plain language standards. However, any such integration must be guided by the human drafter’s judgment; AI suggestions to simplify text should be reviewed to ensure legal precision isn’t lost. The Manual’s section on plain language could be updated to recommend drafters use AI grammar and style checkers as aids for clarity, while cautioning that the drafter must verify and refine AI outputs to maintain accuracy and intent.
Structuring Legislative Drafts: Good legislative drafting isn’t just about words—it’s about structure and logic. Chapter 7 of the Manual, “Structuring Legislative Drafts,” advises that each distinct legal proposition be contained in a single sentence (usually a section or subsection) to maintain unity of purpose. It also provides practical guidance on sentence length and formatting: readers generally “cannot digest more than 4 to 5 lines of unbroken text”, so lengthy provisions should be broken into sub-paragraphs or lists for readability. Conversely, too many short sentences or excessive cross-references (“linking devices”) can hinder understanding. AI tools could assist drafters in adhering to these structural best practices. By updating the Manual to include AI-assisted structuring, drafters could be encouraged to use software that visualises the structure of a bill and highlights deviations from the recommended format. Still, as per the Manual’s ethos, these tools should augment, not replace, the drafter’s own skill in logically organising legislation.
Drafting Procedures and AI Integration: Beyond language and structure, the drafting process itself – from initial policy instructions to final bill – could benefit from AI, and the Manual’s procedural guidance should reflect this. The 2017 Manual provides appendices on the procedure for the preparation of legislation and the content of drafting instructions, outlining how a bill moves from an idea to a Parliament-ready draft. These procedures stress coordination with policymakers and respect for parliamentary rules (for instance, publishing draft Bills with enough lead time for review). Inserting AI into this workflow requires clear protocols. The Manual could be revised to suggest where and how AI tools may be used in the drafting timeline. For example, during the policy analysis stage, AI might compile comparative legislation from other jurisdictions to inform the drafter.
During drafting, AI could generate preliminary text for routine or technical provisions (like definitions or consequential amendments), which the human drafter then curates. However, the Manual should caution that ultimate control rests with human drafters and decision-makers at every step. Under no circumstances should an AI’s draft skip the normal vetting by the sponsoring ministry, legislative counsel, or necessary approvals. If anything, AI’s involvement calls for more transparency: every AI-generated contribution should be documented. The drafting procedure section might include new steps like an AI review checkpoint, where drafters explicitly review AI-suggested text for legal soundness and policy alignment, ensuring nothing enters a bill without human validation.
Recommendations for Updating Legislative Drafting Manuals across the Commonwealth
To responsibly integrate AI into legislative drafting across Commonwealth countries, the following concrete reforms to Legislative Drafting Manuals are proposed:
AI-Assisted Drafting Guidelines: Include a dedicated section on how to use AI tools in legislative drafting. This should offer practical guidance on using AI for tasks like preliminary research, generating draft language, and checking for consistency. It would stress that AI outputs are suggestions to be evaluated – much like how junior counsel’s drafts are reviewed by seniors. Guidelines can cover version control (e.g. keeping track of what portions of text an AI helped produce) and best practices for validating AI contributions against policy intent and legal standards. By establishing formal guidelines, the Secretariat would help drafters use AI’s productivity benefits without compromising quality or clarity.
Principles for Human Oversight: The Manual should lay down principles ensuring that human judgment remains paramount whenever AI is used. A possible principle is: “AI can assist in drafting, but a human legislative counsel is accountable for the final text.” This aligns with expert consensus that human oversight is crucial – as seen in the UAE case, “the UAE’s use of AI to speed up lawmaking… also highlights just how crucial human oversight remains. Experts agree that humans need to keep their hands on the steering wheel.” The Manual can enumerate oversight checkpoints: for instance, requiring that any AI-drafted provisions undergo line-by-line review by a human drafter, and that committees reviewing bills are informed of any AI involvement. Such oversight principles echo the “human-in-the-loop” approach recommended by industry experts to maintain fairness and transparency. They ensure that AI serves as a tool under the direction of human drafters, not an autonomous lawmaker.
Ethical Framework for AI in Law-Making: Legislatures and drafting offices should adopt an ethical framework consistent with broader AI governance norms to guide AI usage. The Manual could reference key values like accountability, transparency, fairness, and respect for human rights in the context of AI-generated legislative text. This would harmonize with international guidelines such as the OECD.AI AI Principles, which urge that AI be “innovative and trustworthy and… respects human rights and democratic values.” Concretely, the Manual might require impact assessments for AI tools (checking for bias or rights implications) and mandate transparency about AI’s role in drafting (so that the origin of legislative language is clear). It could encourage Commonwealth jurisdictions to establish oversight bodies or ethics committees to review the use of AI in their legislative processes. As one expert noted, “Governments, tech firms, legal experts, and ethicists must work together to stay aligned with global norms like the EU AI Act and OECD principles.” An ethical framework section in the Manual would embed these global principles into Commonwealth practice, ensuring AI is used in a manner that bolsters – rather than erodes – the rule of law.
By implementing these reforms, the Commonwealth Secretariat can ensure its drafting manual remains at the cutting edge of best practice. In 2008, Commonwealth Law Ministers recognised the need for updated strategies and welcomed the development of drafting manuals and process guidelines– today, the emergence of AI in law-making is exactly the kind of development that warrants an update. The Commonwealth Legislative Drafting Manual has long been a “landmark tool for promoting the rule of law”; updating it with AI-oriented guidance will help drafters harness new technology while steadfastly preserving democratic integrity.
Conclusion
The UAE’s foray into AI-driven legislative drafting is a wake-up call and an opportunity. It demonstrates the incredible potential of technology to aid in crafting laws – faster analysis, instant data-driven insights, and perhaps a reduction in drudgery – but it also starkly highlights the non-negotiables of democratic law-making: transparency, debate, and human accountability. The Commonwealth, with its diverse member states and shared legal traditions, can take a leadership role by proactively setting standards for ethical AI integration in legislation drafting.
In the spirit of the Commonwealth Secretariat’s commitment to innovation guided by principle, we must ensure that AI serves as a tool of empowerment for legislative drafters, not a replacement for the open, accountable process that turns draft bills into just laws. By updating the Drafting Manual and similar guidelines, drafters will be better equipped to use 21st-century tools to produce laws that are clear, effective, and legitimate. The goal is a future where AI augments human creativity and diligence in lawmaking, without ever bypassing the fundamental democratic processes that safeguard our constitutional freedoms. Each law must still win the test of parliamentary scrutiny and public trust – now with AI as a partner, not a gatekeeper, in that journey.
Sources:
Comments