Burning issue of legal liability when AI causes damage

The Ministry of Justice has just publicly disclosed the Appraisal Dossier for the draft Decree of the Government detailing a number of articles and measures to implement the Law on Artificial Intelligence.

In the Draft Submission, the Ministry of Science and Technology said that the reality of AI deployment may give rise to many problems such as ethical risks, algorithmic biases, privacy infringement, protection of personal data, business secrets and discrimination in AI application; the lack of a mechanism to classify and control AI system risks; lack of a mechanism to inspect, license and supervise AI systems, especially high-risk AI systems; difficulties in tracking responsibility when AI systems cause damage; lack of effective coordination mechanisms in building and sharing common computing infrastructure for AI;...

All of this creates a major barrier to the research, development, deployment and use of artificial intelligence systems in Vietnam.

Functional agencies identify some typical issues such as:

Inadequacies in data management, exploitation and protection - the core resource for AI: Resolution 57-NQ/TW has clearly stated that it is necessary to "put data into the main production material" and "develop the data economy and data market".

However, the reality of collecting, processing, sharing and using data for training AI models is facing many difficulties. Current regulations on personal data protection, although available, are not detailed enough to solve the specific problems of AI (e.g., aggregated data, anonymous data, the right to interpret). The connection and sharing of data between national databases and industry databases is still slow, not meeting the requirements for developing large-scale AI systems.

Notably, the issue of legal liability gaps when AI causes damage: This is one of the most pressing legal challenges from practice. When an AI system causes damage (e.g., self-driving cars cause accidents, medical diagnostic software gives incorrect results, credit systems automatically refuse lending in a discriminatory manner), who will be responsible? Developer, manufacturer, user, or the AI system itself? Current civil and criminal law does not have a satisfactory answer to these situations. This causes confusion for consumers and risks for businesses.

In addition, there are challenges to national security, social order and ethical values: The development of AI poses very urgent non-traditional security risks. Deepfake technology is abused to create false information, defame, deceive or cause socio-political instability,...

Therefore, according to the Ministry of Science and Technology, it is necessary to develop detailed regulations guiding the implementation of the Law on Artificial Intelligence, creating a solid foundation for managing and promoting research, development, deployment and use of artificial intelligence in Vietnam.

Having a comprehensive legal system will be a lever to help Vietnam overcome challenges, build a dynamic AI economy, and affirm the position of a country with technological autonomy.

See the original here