New Regulations Under Consideration to Tackle AI Disinformation and Biases

To counter the spread of disinformation and the biases that accompany artificial intelligence, officials are exploring new regulatory measures that would enable the naming and shaming of AI companies found guilty of significant human rights infringements. While this legislation seeks to enhance accountability, it will not impose criminal penalties or fines, as there are concerns that such punitive measures could impede innovation, according to information from government sources.

This new bill is slated for introduction in February during the ordaining Diet session, where it will endow the government with the power to examine instances where AI-generated content perpetuates discrimination or violates human rights. Under this proposal, businesses would be obliged to assist government inquiries and follow directives, with potential public naming for those that fail to comply. Even if an AI service does not warrant a severe naming penalty, the government plans to inform the public about the service and its peers to raise awareness.

The specific criteria for public naming will be finalized after further deliberation. Additionally, the new law will also involve identifying critical infrastructure impacted by AI and providing guidance to both businesses and the public regarding these technologies. Following a call for such regulations from an expert panel last month, governmental officials have been concentrated on fine-tuning the legislation, particularly focusing on how to handle noncompliant businesses.

While the avoidance of criminal penalties reflects an effort to maintain a balance between fostering AI advancement and addressing potential risks, some experts caution that excessively strict regulations could threaten Japan's position in global competitiveness as well as infringe upon freedoms of speech and expression. They suggest that government intervention may be warranted only when voluntary self-regulation by companies proves ineffective. However, there are also critiques regarding the potential ineffectiveness of the naming and shaming approach, particularly when it comes to addressing individuals or international firms, leading some to argue for more significant penalties to ensure compliance.