OpenAI‘s Sam Altman Says UAE Could Test Potential Global Rules for AI

The United Arab Emirates (UAE) is reportedly positioning itself as a leader in regulating and testing artificial intelligence (AI) technologies.

Sam Altman, CEO of OpenAI, said Tuesday (Feb. 13) during a virtual event that the UAE could serve as a “regulatory sandbox” to experiment with AI and develop global rules to govern its use, Bloomberg reported Tuesday.

Altman emphasized the importance of testing AI technologies in a controlled environment before implementing regulations, according to the report.

The UAE has invested in AI and committed to making it a key policy consideration, the report said. However, its ties with China have raised concerns in the United States.

Emirati AI company G42, which is controlled by UAE National Security Adviser Sheikh Tahnoon bin Zayed Al Nahyan, told Bloomberg that the company would scale back its presence in China to address those concerns, per the report.

G42 has been forging partnerships with leading AI companies, including OpenAI, Microsoft and Cerebras Systems, according to the report.

In addition to discussing AI regulations, Altman also revealed OpenAI’s plans to open-source some of its large-language models and develop tools for poorer nations that cannot afford to develop their own AI systems, per the report.

In August, Mastercard said that it had teamed up with the UAE government to promote AI adoption in that country.

The announcement of this collaboration came as Mastercard opened its newest Center for Advanced AI and Cyber Technology in Dubai, saying the center will develop AI-powered financial crime prevention tools, focus on “securing the digital ecosystem and driving inclusive growth,” and act as a hub for fostering and hiring local AI talent.

Also in August, G42 subsidiary Inception, California-based AI research firm Cerebras and the Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) launched their jointly developed Arabic language-trained AI software.

This large language model (LLM) named Jais was trained over 21 days on a supercomputer co-developed by G42 and Cerebras using a purpose-built dataset of 116 billion Arabic tokens and 279 billion English word tokens designed to capture the complexity and nuance of Arabic, the companies said at the time.