Context: In August, the Korea Fair Trade Commission (KFTC) launched a study into the AI market, with a particular focus on GenAI. The study involved around 50 companies, including those that provide foundation models and computing hardware like AI chips. It mainly focused on general business conditions, product and market status, AI-related transactions by sector, and any experiences of unfair trade practices.
What’s new: The KFTC has published its report on generative AI and competition (December 17, 2024 KFTC GenAI and Competition Report Korean). The study, among other things, warns against the restriction of access to essential components, tying, customer inducement and retention, unfair use of technology, and the improper hiring of personnel. It has suggested, among other proposed changes, that there may be a need to amend Korea’s merger review system, particularly following recent Big Tech partnerships such as Microsoft’s hiring of employees from AI firm Inflection earlier this year (which was subject to scrutiny from several antitrust agencies, including in the UK and Germany).
Direct impact: The KFTC has stated that the report was published to help create a market environment where fair competition can be sustained without hindering the rapid growth of the AI sector. It will also serve as policy material for establishing “competitive order” in this market in the future, it said. This marks the first time that the KFTC has released any sort of policy report, indicating a significant shift away from its historical hierarchical approach.
Wider ramifications: The report comes against the backdrop of increasing regulatory activity around AI and the need to amend and improve laws in South Korea and beyond – both within and out of competition. Antitrust agencies across the world have started issuing opinions and carrying out market studies to make their own assessments of the AI market and its competition concerns – including those in the U.S., the UK, Canada, France, Portugal, India and Japan. Beyond the world of antitrust, a major South Korean cybersecurity firm has warned that the use of GenAI models to conduct cybersecurity attacks will increase in 2025.
The report focused on 50 different companies, but it highlighted six major players that were of particular concern: U.S. tech giants Google, Microsoft, Amazon and Meta, and South Korea’s Naver and Kakao.
Hwijin Choi at Bae, Kim & Lee LLC says Nvidia is another key player it intends to monitor closely. The report mentioned that there is a continued excess demand for the company’s GPUs in the AI semiconductor market, which is essential for the GenAI ecosystem, he says. But large-scale operators and vertically integrated businesses that could gain a competitive advantage and create difficulties for others in the market should also be cautious, he adds.
The KFTC proposed two major changes that it believes are worth considering to help address its concerns:
- Reviewing and amending Korea’s merger control regime.
- Closer surveillance of non-obvious practices that may restrict competition, such as the exploitation of data collection.
Sangyun Lee, a postdoctoral research fellow at Kyoto University, says the KFTC’s report exercises special caution regarding the possible expansion of global Big Tech’s market power. Aside from Microsoft’s hiring from Inflection, it also mentions partnerships between Microsoft and OpenAI, Google and Anthropic and Amazon and Adept (in which Amazon hired key personnel from Adept as well as acquiring its AI model license). The report states that these partnerships may need to be reviewed if they have effects equivalent to traditional mergers, he says.
Mr. Lee adds that Korea’s competition law includes a provision addressing unfair recruitment practices that hinder the activities of other businesses as a type of unfair trading practice – and the possibility of applying this framework is also noted in the report.
“However, while it may be prudent to amend the merger review system to meet specific needs, amending it without thorough study and review could impose burdens on businesses,” Mr. Choi warns.
On data collection, the KFTC says it plans to explore reforms to regulate actions by AI operators related to data collection and usage that may harm consumer interests. This suggests obtaining genuine consent for data collection would be advisable, Mr. Choi says.
Today’s KFTC report is the first time the authority has published a policy paper. This marks a “significant and positive shift away from its previous hierarchical approach”, according to Mr. Lee.
Until this week, the Korean competition watchdog had taken a much more passive approach to regulation. It has previously attempted to introduce an ex-ante platform regulation but these efforts faced significant backlash and failed to continue.
This backlash, according to Mr. Lee, was largely because the initiatives were implemented by the government in a very “hierarchical top-down manner”, without public consensus regarding the competition issues in Korean platform markets. This is starkly different to the processes adopted by Australia, the EU, its member states and Japan, which have issued policy papers as part of a “broader consensus-building effort”, he says.
Mr. Lee believes this approach is an “essential first step” toward making rational policy decisions in the future:
“The KFTC’s approach taken this time on AI issues—focusing on identifying the current situation and potential concerns—is commendable. The report systematically examines market conditions, identifies potential problem areas, and evaluates whether existing tools are sufficient to address these issues or if there are gaps that need to be filled.”
Wider AI concerns raised
Korea’s competition watchdog is not alone in its concerns surrounding competition and GenAI.
In fact, this year alone, the UK’s Competition and Markets Authority (April 11, 2024 CMA press release), France’s Autorité de la Concurrence (June 28, 2024 Autorité press release), Canada’s Competition Bureau (March 20, 2024 Competition Bureau report) and the Organisation for Economic Co-operation and Development (May 24, 2024 OECD report) have all published reports or detailed opinions on competition in the AI sector. Portugal’s Autoridade da Concorrência also issued a general warning of competition concerns in the AI industry last year (November 6, 2023 AdC press release).
In July, the European Commission issued a joint statement with the CMA, the U.S. Federal Trade Commission and the U.S. Department of Justice highlighting the risks involved in this market – such as how some platforms may have substantial market power at multiple levels related to the AI stack – and the ways that they plan to protect competition, such as closely scrutinizing “claims that interoperability requires sacrifices to privacy and security” (July 23, 2024 ai fray article).
Meanwhile, three other agencies are currently in the middle of conducting their own studies, including the U.S. FTC (January 25, 2024 FTC press release), the Competition Commission of India (April 22, 2024 CCI proposal) and Japan’s Fair Trade Commission (October 2, 2024 JFTC press release).
The KFTC’s study comes the same week as a report published by South Korean cybersecurity firm SK Shieldus warning that hacking groups are expected to increasingly use GenAI models, such as ChatGPT, to conduct more cyberattacks in 2025. AI will be leveraged to create spear phishing emails customized to their attack subjects and fake news materials to be used for political propaganda, the report warned, and “it will be difficult to tell the authenticity of sophisticated content created with the help of AI”.
According to an annual study conducted by Verizon, there were 2,130 cyberattacks in the Asia-Pacific region this year (2024 Verizon report). While North America and EMEA witnessed much higher numbers – 16,619 and 8,302, respectively – APAC saw the highest level of secrets being compromised due to cyberattacks, with 24% relating to such breaches. The report also warned of a possible increase in blackmail cases using deepfake materials – South Korea saw a series of deepfake sex crime cases in early 2024, with police having arrested 506 suspects in such cases as of October (September 26, 2024 Korea Times article) – and state-sponsored hacking groups due to global political uncertainties following the re-election of Donald Trump as U.S. President. This is because the U.S. is expected to make policy shifts toward easing regulations on technology and virtual asset industries, and promoting protectionism, the report stated.