On July 15, 2024, the Taiwanese National Science and Technology Council released a draft of the Artificial Intelligence Basic Act, revealing seven basic principles, including sustainable development, human autonomy, privacy protection, security and safety, transparency and explainability, fairness and non-discrimination, and accountability, as the directions for Taiwan to enhance the development of artificial intelligence ("AI") technology and applications. Article 9 of the draft provides: "The government shall prevent the applications of AI from causing damages to citizens' lives, bodies, freedom or property safety, social order, and the ecological environment, or from violating laws and regulations by incurring conflicts of interest, bias, discrimination, false advertisement, misinformation and disinformation, etc." Among other things, the issues of discrimination, false advertisement, misinformation and disinformation essentially involve Competition Law. Here, the draft also specifically states: the Taiwanese Fair Trade Act shall be one of the laws that might be applied to the applications of AI. This article will explore the unfair competition issues caused by the improper applications of AI from two aspects: "false advertisement" and "price discrimination" that AI may involve. 

AI-generated advertisements may be misleading and false
The groundbreaking development of AI lies in the machines' ability to learn on their own. Computer systems can learn and make decisions from massive amounts of data, instead of passively receiving explicit human orders or program codes to operate as in the past. The industries are also endeavoring to exploit this emerging technology in the development of advertisement generators in order to generate advertisement materials including texts and images at lower costs and higher efficiency. This year, however, the Taiwanese Fair Trade Commission ("FTC") punished the keyword advertisement of the well-known hotel booking website Agoda. This case reminds us that if a review mechanism for "self-learning machine" is not properly established, the business will face the risks of heavy fines under Taiwanese laws. 

Agoda was fined because of using its competitor's name, Easytravel, in the keyword advertisement, and showing "Easytravel/Late birds also enjoy special discounts: Up to 80% off" and Agoda's domain name in the title of advertisement. FTC found that the advertisement as a whole may easily mislead internet users into thinking that Agoda and its competitor, Easytravel, are from the same source or are related companies, thus constituting an unfair action of exploiting others' efforts. Pursuant to the decision, Agoda purchased keyword advertisement in the following process: Agoda registered an account on the keyword advertisement platform for purchasing keyword advertisement, and then connected Agoda's "self-learning machine" with the search engine system through the "application programming interface" (API) of the said platform. So, the "self-learning machine" would manage the account and purchase keyword advertisement automatically. From the path taken by internet users to visit the Agoda website, Agoda learned the keywords input by any specific user in the search engine and collected the keywords of all internet users on a daily basis for the "self-learning machine" to decide which keywords to purchase and post. Based on Agoda's statements given to the FTC, the "self-learning machine" recommended huge number of keywords, Agoda had never sent any employee to manage the purchase of keywords, nor appointed any full-time personnel to review the purchased keywords. Thus, whether you can effectively monitor the legal risks of AI certainly is a key for the internal control and audit system of businesses, when taking advantage of the convenience brought about by AI. 

Personalized pricing involves issues of price discrimination
Racial or sex discrimination are mostly discussed when it comes to discrimination issues relating to AI. In terms of competition law, price discrimination is an important issue worthy of attention. Price discrimination is also known as “differential pricing”, which means that an enterprise sets different prices for different trading groups or counterparties. Ultimately, an enterprise may charge each person for different prices, which is known as personalized pricing. Personalized pricing will be more realizable given that AI collects and analyzes big data and then becomes increasingly able to grasp consumers' habits and preferences, thereby accurately calculating the price sensitivity of individual consumers. The OECD once issued the report "Personalised Pricing in the Digital Era", pointing out that personalized pricing is a kind of price discrimination, which in some circumstances may cause harm to consumers and needs to be addressed through a combination of policy tools, including competition policy. The U.S. FTC further ordered eight enterprises, including MasterCard, on July 23, 2024, for production of information, as these enterprises claimed that they had utilized AI or other technologies in obtaining historical or real-time consumers' information for the purpose of individual consumer pricing. Obviously, the U.S. FTC already focused on the issue of personalized pricing and is taking its initiative on it. 

In Taiwan, price discrimination is mainly regulated according to Subparagraph 2 of Article 20 of the Taiwanese Fair Trade Act "treating another enterprise discriminatively without justification". However, it only applies to the "business to business" discrimination, not applicable to the "business to consumers" discrimination, based on the literal interpretation and practice of this subparagraph. Thus, the law currently in force cannot be properly applied to the situation wherein AI collects consumer information and performs personalized pricing. Therefore, Article 17 of the draft of the Artificial Intelligence Basic Act requires the Ministries to review current laws, regulations, and measures, and make amendments in accordance with the gist of this Act to cope with future changes in the AI era.