On April 14, the leader of Dubai, Sheikh Mohammed bin Rashid Al Maktoum, declared that the United Arab Emirates would commence employing artificial intelligence to assist in legislative drafting. A new Regulatory Intelligence Office would leverage the technology to “regularly propose updates” to the law and “expedite the issuance of legislation by up to 70%.” AI would formulate a “comprehensive legislative strategy” encompassing local and national laws and would be interconnected with public administration, the judiciary, and international policy developments.
The initiative was met with widespread disbelief. This type of AI governance could be a global “first,” with the possibility of going “astonishingly wrong.” Critics worry that the AI model may fabricate facts or drastically misinterpret societal principles like equity and justice when shaping legislation.
In reality, the UAE’s concept of AI-generated law is not genuinely unprecedented and isn’t necessarily problematic.
The earliest instance of enacted law believed to have been authored by AI was passed in Porto Alegre, Brazil, in 2023. It concerned a local statute regarding water meter replacement. Council member Ramiro Rosário sought assistance in generating and articulating solutions for a policy challenge, and ChatGPT performed sufficiently well for the bill to pass unanimously. We endorse the use of AI in assisting humans in this capacity, although Rosário should have revealed that the bill was drafted by AI prior to the vote.
Brazil was a precursor but far from singular. In recent years, a continual flow of ambitious politicians at both local and national levels have introduced legislation that they advertise as being designed by AI or allowing AI to formulate their speeches or even deliver them in sessions.
The Emirati initiative differs from these cases in significant ways. It aspires to be more systematic and not merely a publicity stunt. The UAE has committed to investing over $3 billion to evolve into an “AI-native” government by 2027. Only time will reveal if it is also distinct in being more about publicity than substance.
Instead of being a genuine first, the UAE’s revelation reflects a much broader global movement of legislative bodies embracing AI assistive tools for legislative research, drafting, translation, data analysis, and beyond. Individual lawmakers have begun utilizing AI drafting tools as they traditionally relied on aides, interns, or lobbyists. The French government has even taken steps to train its own AI model for legislative tasks.
Even requesting AI to thoroughly review and revise legislation would not be a novel occurrence. In 2020, the U.S. state of Ohio started employing AI for extensive revisions of its administrative law. The speed of AI can be a suitable fit for such large-scale editing endeavors; the state’s former lieutenant governor, Jon Husted, claimed it effectively removed 2.2 million words of unnecessary regulation from Ohio’s statutes. Now a U.S. senator, Husted has recently recommended a similar methodology for U.S. federal law, advocating for AI as a means for systematic deregulation.
The risks of misinformation and lack of humanity—while valid—are not what makes the potential of AI-generated legislation groundbreaking. Humans also make errors when drafting laws. Recall that a single typo in a 900-page statute nearly undermined the massive U.S. health care reforms of the Affordable Care Act in 2015, before the Supreme Court overlooked the error. Moreover, distressingly, citizens and inhabitants of nondemocratic regimes are already subjected to arbitrary and often brutal laws. (The UAE is a federation of monarchies without direct legislative elections and presents a poor record on political rights and civil liberties, as assessed by Freedom House.)
The primary worry regarding the use of AI in legislation is that it will serve as a tool for the powerful to promote their own agendas. AI may not radically transform lawmaking, but its extraordinary capabilities could intensify the dangers of power concentration.
AI, and technology in general, is frequently invoked by politicians to lend their projects an air of objectivity and rationality, but it doesn’t truly accomplish that. As presented, AI would merely provide the UAE’s hereditary leaders with new resources to articulate, enact, and enforce their favored policies.
Mohammed’s assertion that a main advantage of AI will be to expedite the lawmaking process is also misplaced. The machine may generate the text, but humans will still propose, debate, and vote on the legislation. Drafting is seldom the hurdle in enacting new laws. What consumes far more time is for humans to amend, negotiate, and ultimately forge consensus on the substance of that legislation—even when that bargaining occurs among a select group of monarchical elites.
Beyond speed, the more crucial capability offered by AI is intricacy. AI has the potential to render laws more nuanced, tailoring them to a multitude of specific scenarios. The fusion of AI’s research and drafting efficiency enables it to outline legislation governing an array of special circumstances for each proposed rule.
However, this aspect of AI once more allows the powerful to assert their will. AI’s ability to compose intricate laws would enable the individuals guiding it to impose their precise policy preferences for each unique case. It could even covertly embed those preferences.
Throughout history, legislators have crafted legal loopholes to specifically favor special interests. AI will become a potent instrument for authoritarian regimes, lobbyists, and other influential interests to accomplish this on a larger scale. AI can facilitate the automatic creation of what political scientist Amy McKay has termed “microlegislation”: loopholes that may go unnoticed by human readers on the page—until their consequences manifest in reality.
Nevertheless, AI can be regulated and directed to disseminate power rather than consolidate it. For Emirati residents, the most fascinating potential of the AI initiative is the promise of introducing AI “interactive platforms” where the public can contribute input on legislation. In trials across varied locations such as Kentucky, Massachusetts, France, Scotland, Taiwan, and many others, civil society in democracies is innovating and testing ways to utilize AI to better engage with constituents and shape public policy that serves a broad array of stakeholders.
If the UAE intends to develop an AI-native government, it should pursue this goal to empower citizens rather than machines. AI possesses substantial potential to enhance deliberation and pluralism in policymaking, and Emirati residents should hold their government accountable to fulfilling this promise.