UAE to use AI to write laws – A historic turning point in world lawmaking

The UAE is moving towards a revolution in the legislative sector, announcing that it will use AI to help draft new laws and review and edit existing ones.

This is considered the Gulf nation's boldest step yet in its journey to fully exploit AI technology, an area in which the UAE has invested tens of billions of dollars in recent years.

According to AI experts, the UAE's “AI-powered legislation” initiative goes beyond any similar efforts undertaken in other countries, although details are still limited.

While many governments around the world are exploring the possibility of applying AI to summarize legal documents or improve the efficiency of public service delivery, the UAE is proactively using this technology to propose law amendments, based on analysis of legal and administrative data.

“The new AI-powered legislative system will completely change the way we make laws, making the process faster, more accurate and more practical,” Sheikh Mohammad bin Rashid Al Maktoum, Vice President and Prime Minister of the UAE, told state media.

Last week, the UAE approved the establishment of the Office of Legal Intelligence – a new cabinet unit tasked with overseeing the integration of AI into legislative activity.

Professor Rony Medaglia from Copenhagen Business School (Denmark) assessed this as an ambition to "turn AI into a legislative tool", and called this move "very bold".

In addition, the UAE continues to boost investments in AI through the $30 billion MGX technology investment fund, in partnership with the US asset management group BlackRock. Notably, MGX also appointed an “AI observer” to the board of directors, a rare move globally.

In the future, the UAE will use AI to analyze the impact of laws on society and the economy by building a massive data warehouse that integrates federal and local laws, court judgments, and data from government agencies. This promises not only a more flexible and efficient legal system, but also opens a new era in technology-driven national governance.

According to state media, Sheikh Mohammad said AI will “regularly propose updates to our legal system.” The UAE government expects AI to speed up legislation by up to 70%, according to a cabinet meeting announcement.

However, researchers warn that AI can face many challenges and pitfalls. These range from AI becoming difficult for users to understand, to its training data being biased, and questions about whether AI can interpret laws the same way humans can.

While current AI models are impressive, “they can still be delusional and have problems with reliability and stability,” warns Oxford University (UK) researcher Vincent Straub. “We cannot fully trust them.”

Mr. Straub also emphasized that the UAE’s plan is unique because it uses AI not simply to improve existing laws, but also to predict future legal changes that may be needed. This could save money, as governments often have to pay law firms to review and evaluate regulations.

“It seems like they're taking it a step further… from looking at AI as a tool to assist, sort, and draft, to using AI to predict and plan for changes,” Straub said.

Keegan McBride, a lecturer at the Oxford Internet Institute, said that while democratic countries may struggle to adopt digital government, the UAE, with its authoritarian political system, has been able to move quickly and experiment with new initiatives. “They can make rapid changes and test multiple models,” he said.

McBride said that while there are dozens of small ways other countries are using AI in the legal sector, he had not seen a similar plan. “In terms of ambition, the UAE is currently at the top.”

It is unclear at this point which AI system the UAE government will use, and experts say it may need to be a combination of different systems. However, researchers agree that setting up barriers and human oversight of AI will be important.

In particular, Marina De Vos, a computer scientist at the University of Bath (UK), warned that AI can make “really strange” suggestions that machines can understand, but are not suitable when applied to human social practice.