Dear Minister Nguyen Manh Hung,

On behalf of the Center for Data Innovation (datainnovation.org), I am pleased to submit these comments to the Ministry of Information and Communications on Vietnam’s draft Law on Digital Technology Industry (DTI).[1]

The Center for Data Innovation studies the intersection of data, technology, and public policy, and formulates and promotes pragmatic public policies designed to maximize the benefits of data-driven innovation in the public and private sectors. It educates policymakers and the public about the opportunities and challenges associated with data, as well as technology trends such as open data, artificial intelligence, and the Internet of Things. The Center is part of the Information Technology and Innovation Foundation (ITIF), a nonprofit, nonpartisan think tank.

Overall, the draft DTI law provides a welcome policy framework for supporting Vietnam’s development and adoption of key technologies, including semiconductors, cloud computing, the Internet of Things, blockchain, augmented and virtual reality (AR/VR), and artificial intelligence (AI). The framework includes creating incentives for firms to invest in digital technologies, establishing regional technology parks, developing 5-year plans for boosting smart industries (e.g. smart agriculture, smart manufacturing, etc.), training and attracting workers with digital skills, and encouraging international cooperation and engagement on digital policy. Notably, the draft DTI law emphasizes improving cooperation and collaboration between universities, research institutions, startups, and established firms to create a robust digital innovation ecosystem. While the relevant government ministries still need to provide appropriate and relevant details about each of these policies, creating a comprehensive set of government policies to support Vietnam’s digital technology sector will boost the nation’s innovation capacity.

In addition, the draft DTI law outlines new policies around the management of data and the use of AI. With regards to data, the draft law emphasizes the importance of fostering the accessibility and availability of data. For example, the law calls for government agencies to allow individuals and organizations to access open data. While the Ministry of Information and Communications has already established an open data portal and has made significant progress on making some national databases available to the public, such as the national population database and the national database on business regulation, it has not yet released other datasets, such as those on finance, population statistics, and criminal justice.[2] Releasing more open data will allow organizations and individuals to put this data to productive uses.

The draft DTI law also calls for some data portability requirements. The law would require businesses to allow customers to transfer data from their products and services to the same types of products and services offered by competitors. In addition, businesses would be prohibited from imposing technical or legal restrictions that restrict data portability. Generally, data portability is a useful policy that can increase consumer welfare and foster innovative uses of data.[3] However, the proposal in the draft law limits the data portability obligations only to moving data to the same types of products and services. For example, this data portability requirements would allow consumers to move their data from one fitness tracking device to another, but it may not allow consumers to import their data from the fitness tracking device into a fitness mobile app. In some cases, this requirement might be too narrow to foster useful data-driven innovations. It would be better to allow customers to transfer their data for any purpose.

Finally, the draft DTI law includes several measures to support the creation of high-quality datasets including creating sample data sets to support digital technology development, data deidentification, and the development of data quality assurance measures. Each of these measures can contribute to greater availability and accessibility of high-quality datasets. In addition, the Ministry should create guidelines around additional organizational models to facilitate data sharing, such as data consortia (where organizations pool their data together), data trusts (where entities delegate their data rights to an independent intermediary), and data cooperatives (where entities negotiate collectively for access to their data).[4]

With regards to AI, the draft law provides a risk-based framework for the safe and responsible development and deployment of AI. As part of this framework, the draft law prohibits certain types of AI systems, such as those used to significantly harm individuals by subconsciously manipulating their behavior, discriminating against certain groups based on social behavior (i.e., social credit scoring), using AI for mass surveillance with facial recognition, and inferring human emotions in the workplace or in educational settings. Many of these prohibitions focus too narrowly on AI. For example, presumably the government would desire to prohibit any product or service—whether or not it involves AI—that causes significant harm to individuals by exploiting aspects of their disabilities, age, or economic circumstances. In these cases, the law should not target AI specifically and aim to create technology-neutral policies. In other cases, the draft DTI law is too broad, such as prohibiting AI systems to infer human emotions in certain settings, without considering that there may be useful applications of this technology, such as to better personalize AI-enabled interactions based on the mood of the individual.

Many of the details of the AI regulations in the draft DTI law will be determined at a later date by the Ministry of Information and Communications, such as the relevant risk levels for different categories of AI systems and the associated responsibilities and obligations at each level. As the Ministry proceeds with classifications, it should aim to provide sector-specific classifications rather than broad, horizontal classifications for AI systems. For example, an AI system used in the financial services sector may be quite different from one used in the transportation sector or the healthcare sector. Classifying AI systems based on specific sectors can help avoid imposing inappropriate and unnecessary obligations that would make developing and deploying this technology more expensive.

Finally, the draft DTI law proposes to require businesses to label the output of AI systems in a machine-readable and detectable format. While there are some methods that can achieve this goal, such as adding metadata to files or adding watermarks to the content of the output itself, users can easily remove these signals, thereby limiting their effectiveness. Moreover, some types of output, such as text, are much more difficult to reliably label, and ultimately may prove to be a technological dead end.[5] Given the limitations of labelling, the Ministry should drop these requirements until global standards bodies develop robust labelling methods for specific types of digital media.

To conclude, the draft DTI Law presents a promising framework for advancing Vietnam's digital technology sector. While it offers valuable steps forward in areas such as data accessibility, AI regulation, and industry support, there are several aspects that could benefit from refinement. These include:

▪ Broadening the data portability requirements to allow more innovative data uses.

▪ Creating technology-neutral AI policies and avoiding overly broad AI prohibitions.

▪ Establishing sector-specific classifications for AI risk levels.

▪ Eliminating the AI output labeling requirements given the nascent state of the technology.

By addressing these points, the DTI law can better foster innovation while maintaining responsible development and use of digital technologies. We recommend continued collaboration with stakeholders to fine-tune these policies, ensuring they effectively support Vietnam's digital transformation while addressing potential risks and challenges.

Endnotes