Saudi Arabia Enacts Landmark Copyright Law Permitting AI Model Training
Riyadh, KSA – Saudi Arabia has introduced a significant legal development in the realm of intellectual property, with a new law now explicitly permitting the use of copyrighted materials for the training of Artificial Intelligence (AI) models. The legislation, which came into effect in August, positions the Kingdom alongside global leaders like Japan, the European Union, and Singapore in establishing a clear legal framework for AI development in relation to copyright.
This move underscores Saudi Arabia’s strategic ambition to emerge as a key player in the global AI landscape, fostering both computational power and the development of indigenous AI models and tools. The new framework is anticipated to be particularly beneficial for advancing Arabic-language AI capabilities.
Key Provisions and Implications
Under the new law, developers operating within Saudi Arabia are authorized to train AI models using copyrighted works, provided that the materials were obtained lawfully and are utilized exclusively for model training purposes. A crucial requirement for compliance dictates that developers must meticulously log the type, source, purpose, and date of each use.
The legislation also clarifies that while copyrighted content can be used for training, the AI models themselves are not permitted to reproduce the original works after training. This distinction aligns with the principle that AI learning should parallel human learning, where knowledge acquisition does not equate to verbatim reproduction.
A notable aspect of the implementing regulations supporting the law facilitates the translation of works from other languages into Arabic by Saudi entities. This is permissible without the explicit consent of the copyright holder if an Arabic translation has not appeared within a specified period, typically three years. While not solely aimed at AI, this provision could significantly increase the volume of Arabic training content available for AI models across various dialects. Legal experts believe this will accelerate the development and prosperity of Arabic-language AI models.
International Context and GCC Comparison
Saudi Arabia’s proactive approach to AI copyright diverges from practices in some other jurisdictions. In the United States, for instance, AI companies often rely on the “fair use” exception to copyright law, a doctrine that remains a subject of ongoing legal debate and lawsuits from copyright holders, including prominent media organizations.
The trend globally appears to be moving towards establishing clear carve-outs or interpretations of existing laws that permit AI training on copyrighted materials, balancing innovation with accountability.
In the United Arab Emirates, current copyright law does not feature a specific carve-out for AI model training or a comprehensive “fair use” doctrine. While this has not yet led to significant legal challenges for the UAE’s burgeoning AI ecosystem, a flagship Arabic AI model developed in the UAE was notably trained on servers located in California, thereby falling under US copyright protections. Industry observers suggest that given the benefits of a specific legal framework, the UAE may consider updating its copyright regime in the near future, potentially within the next 18 months, to better support its AI ambitions.
Future Considerations and Safeguards
Despite the forward-looking nature of the Saudi law, international legal complexities persist. The global nature of AI models means that legal regimes vary significantly across countries, potentially creating challenges for models trained under one jurisdiction when deployed in another. For example, provisions within the European Union’s AI Act could impose compliance requirements on general-purpose AI models operating in the EU, regardless of their training location, particularly concerning opt-out mechanisms for copyright holders.
Legal specialists emphasize the ongoing tension between intellectual property rights and AI innovation. They advise Saudi technology companies to implement robust safeguards, such as ensuring models do not retain data post-training and offering clear mechanisms for copyright holders to opt out of their content being used for AI training. Ultimately, the immediate priority for AI developers remains focused on building competitive models, securing talent, and ensuring necessary infrastructure.