The US FTC's AI Risk Investigation and the "AI-to-SI" Nomenclature Debate: US-China AI Competition and Implications for South Korea
Executive Summary
The US FTC's investigation into OpenAI and Anthropic, the White House's executive order renaming "AI" to "SI," and the agreement on a superintelligence dialogue at the US-China summit are separate tracks originating from different institutional logics. While the White House prioritizes external posturing and domestic political framing, the FTC operates independently based on consumer protection jurisprudence, resulting in a gap in the consistency of US AI policy. The fact that President Trump reaffirmed his stance against sharing AI governance with China even immediately after the summit suggests that the SI dialogue is likely to remain symbolic, a view supported by China's decision not to officially adopt the term "SI." South Korea needs to respond by decoupling security-linked technology controls from commercial AI cooperation, and separately track the FTC's investigation trends from the perspective of managing risks for South Korean companies entering the US market. Rather than preemptively responding to the name change, South Korea should pursue a strategy of securing a voice in the formation of multilateral norms that are not subordinate to the bilateral US-China dynamic.
I. Issue Analysis
US FTC's AI Risk Investigation and the "AI-to-SI" Nomenclature Debate: Issue Analysis
1. Background and Developments
Washington's AI policy landscape reached a turning point with the Trump-Xi summit on September 24–25. This summit did not occur out of the blue. During their Beijing summit in May, the two leaders had already agreed in principle to initiate an intergovernmental dialogue on AI [3][9]. The September summit in Washington served as a follow-up to this agreement, coinciding with the symbolic significance of President Xi's first visit to the White House in 11 years [3][9].
In a fact sheet released immediately after the summit, the White House announced that the two countries had established a "US-China Super Intelligence (SI) Dialogue" [1]. A communication channel for mutual notification in the event of AI accidents was also newly established [1][9]. It was agreed that the next meeting would be held by November [1][6]. However, the term "superintelligence" has not been officially adopted by the Chinese side [1]. An EAI analysis points out that this term was "first established by President Trump as a domestic political discourse on Truth Social prior to his UN General Assembly speech" [3]. In other words, a domestic political frame preceded the diplomatic agreement.
This trend entered the institutionalization phase in October. According to reports by local media in Oman, President Trump signed an executive order directing all sectors of the federal government to officially replace the terms "Artificial Intelligence" and "AI" with "Super Intelligence" and "SI" [7]. The executive order is titled "Inaugurating the Era of Super Intelligence" [7]. The terminology change has thus been officially reflected in the federal administrative system, moving beyond mere rhetoric.
Around the same time, the FTC was moving on a separate track. According to a Hankyoreh article citing a New York Post report on September 30, a senior FTC official stated that "Chairman Andrew Ferguson initiated investigations into major artificial intelligence companies a few weeks ago" [4]. The targets of the investigation include multiple AI companies, including OpenAI and Anthropic, and the scope of the investigation is set to cover potential violations of the Federal Trade Commission Act [4]. Brazil's O Globo reported that this investigation is being conducted "amid large-scale cybersecurity incidents" [16]. Sweden's Dagens Industri and Denmark's Børsen similarly reported that the FTC is targeting the "safety" (säkerhet) of AI systems [17][19].
2. Current Situation
Currently, Washington's AI policy is operating on two separate tracks. An EAI analysis diagnoses that "the White House track, which signals cooperation, and the regulatory track, which continues investigations into advanced chip export controls and model distillation, are operating separately, creating a structure where symbolic agreements between leaders cannot prevent the continuation of individual regulations against China" [3]. Indeed, immediately after the summit, President Trump reaffirmed his stance that the US would not share AI governance with China [6]. The frames of "achieving superintelligence" and "winner-take-all" remain the official policy stance [6].
The Council on Foreign Relations (CFR) evaluated the summit as demonstrating the "limits of photo-op summits," pointing out that President Trump appeared more focused on getting China to agree to rename AI as "superintelligence" than on substantive safety norms [5]. The Nihon Keizai Shimbun (Nikkei Asia) reported that skepticism exists even within Silicon Valley regarding the hotline with Beijing proposed by Washington [12]. At the White House state dinner, Nvidia CEO Jensen Huang, AMD CEO Lisa Su, Elon Musk, and Apple CEO Tim Cook sat at the same table as Trump and Xi [12], showing that, independent of intergovernmental agreements, the interests of Big Tech companies are directly intertwined with this negotiation table.
China's response remains cautious. According to an EAI analysis, China is "insisting on the sharing of authority to define safety and the application of identical standards to US companies as prerequisites" [6]. In other words, Beijing refuses to unilaterally accept the concept of "safety" defined by the United States. Nikkei described this as a fundamental gap between Xi Jinping's principle of "human control" and Trump's drive to develop "superintelligence" [12]. The structure makes it difficult to find substantive common ground.
The FTC investigation is unfolding on a separate domestic regulatory track. CFR reported that after Anthropic CEO Dario Amodei published an essay urging leading labs to slow down the pace of AI development, figures ranging from Senator Bernie Sanders to President Trump, the Chinese Communist Party, and major corporate CEOs joined the debate [13]. The debate over AI risks appears to have expanded beyond Silicon Valley into the broader political arena.
3. Key Actors and Positions
The White House / President Trumpis sending contradictory signals by simultaneously pushing for the introduction of the term "superintelligence" and the FTC investigation. Externally, the administration projects an image of cooperation by establishing a dialogue with China, while simultaneously using an executive order to officially declare a "winner-take-all" development race [6][7]. EAI evaluates this choice of terminology as a political expression that "places greater emphasis on development than on technology control" [1].
The FTC (Federal Trade Commission)is conducting independent investigations into OpenAI and Anthropic under the leadership of Chairman Andrew Ferguson [4]. This is a separate track operating within the domestic administrative domain of consumer protection and unfair practices regulation, independent of the White House's external message of cooperation.
The Chinese Governmenthas agreed to participate in the dialogue without officially adopting the term "superintelligence," while setting the sharing of authority to establish safety standards and the application of identical standards to US companies as prerequisites [6]. President Xi Jinping's principle of "human control" differs in tone from Trump's development drive [12].
Silicon Valley Big Tech (Nvidia, AMD, OpenAI, Anthropic, etc.)occupies a dual position: acting as practical stakeholders in the US-China negotiations, as evidenced by the seating arrangement at the state dinner, while simultaneously being the targets of the FTC investigation [4][12]. Amodei's essay urging a slowdown shows that risk perception is not uniform even within the industry [13].
4. Key Issues
The first issue is whether the introduction of the term "superintelligence" represents a substantive agreement on norms or merely domestic political rhetoric. Given that China refuses to use the term officially [1][5], this is closer to unilateral framing by the Trump administration rather than a bilateral agreement.
The second issue is the credibility of US policy, which arises from the separate operation of the White House's external cooperation track and the regulatory tracks of the FTC and the Department of Commerce. The establishment of a symbolic dialogue is highly unlikely to lead to an actual easing of export controls or corporate investigations [3][6].
The third issue is the leadership in setting safety standards. China's demand to "share the authority to define safety" is a challenge to the US-led discourse on AI safety, and whether a compromise can be reached on this point at the follow-up meeting in November is expected to serve as a litmus test for the dialogue's effectiveness [9].
This dynamic also has implications for designing technology cooperation within the South Korea-US alliance. Given that the US stance on export controls against China is highly likely to continue regardless of symbolic agreements between the leaders, South Korea needs to maintain an independent management system that does not link semiconductor and AI supply chain risks to announcements from US-China bilateral channels.
II. In-Depth Analysis
US FTC's AI Risk Investigation and the "AI-to-SI" Nomenclature Debate: In-Depth Analysis
1. Root Cause Analysis
The primary cause of this issue lies in the fact that the two pillars governing AI policy within the US administration operate under different logics. The White House prioritizes industrial promotion and external posturing. The FTC operates on separate legal grounds of consumer protection and market fairness. The timing of FTC Chairman Andrew Ferguson's statement that the commission "initiated investigations into major artificial intelligence companies a few weeks ago" almost overlaps with President Trump's signing of the SI executive order [4]. The fact that both actions occurred around the same time in itself demonstrates a lack of policy consistency within the administration.
The roots of the SI renaming debate run deeper. An EAI analysis points out that "the introduction of the term 'superintelligence' was also first established by President Trump as a domestic political discourse on Truth Social prior to his UN General Assembly speech" [3]. This means that rather than being a product of diplomatic negotiations, a domestic political frame existed first, and the summit merely served to endorse it. China has not officially adopted this term [1]. The reaction of local experts reported by Denmark's Børsen also reflects a skeptical view that "this concept will not actually take root" [15]. This indicates that the name change remains at a rhetorical level without accompanying substantive norm changes.
The impetus for the FTC investigation is the accumulated concern over the product safety of AI companies. Brazil's O Globo reported that the investigation is being conducted "amid large-scale cybersecurity incidents" [16]. Sweden's Dagens Industri and Denmark's Børsen also reported that the focus of the investigation is on the "safety" of AI systems [17][19]. Furthermore, CFR evaluated that after Anthropic CEO Dario Amodei published an essay urging leading labs to slow down the pace of development, "the slow-boiling crisis over AI risk came to a head last week" [13]. In a phase where safety controversies are being publicized even within the industry, the justification for the FTC to act as a regulatory body has been strengthened.
2. Structural Context
From a political structure perspective, the Trump administration's AI policy is difficult to explain in isolation from the domestic political calendar. Ahead of the November midterm elections, the frame of "achieving superintelligence" is a political asset that gives voters the impression that the United States is leading the tech hegemony race [6]. At the same time, the FTC, as an independent regulatory agency, conducts its investigations based on consumer protection jurisprudence, regardless of the election schedule. While the separation of these two tracks is institutionally natural, externally it obscures whether US AI policy is a coherent national strategy or a series of sporadic responses by individual agencies.
In the economic structure, the interests of Big Tech companies are intricately intertwined with government policy. The fact that Nvidia CEO Jensen Huang, AMD CEO Lisa Su, Tesla CEO Elon Musk, and Apple CEO Tim Cook sat at the same table as Trump and Xi at the White House state dinner symbolically illustrates this point [12]. The Nihon Keizai Shimbun reports that "the hotline with Beijing proposed by Washington is facing skeptical reactions in Silicon Valley" [12]. While companies welcome the creation of intergovernmental dialogue channels, they remain wary of the possibility that such channels might restrict their technological edge or market access. The inclusion of OpenAI and Anthropic in the FTC investigation is also not unrelated to the fact that they are leading companies in the generative AI market.
In the security structure, a contradiction emerges as both the US and China simultaneously pursue the conflicting goals of "competition" and "management" regarding AI. An EAI analysis diagnoses that "immediately after the summit, President Trump confirmed the frames of 'achieving superintelligence' and 'winner-take-all' as the official policy stance by stating that the US would not share AI governance with China, while China is adhering to the sharing of authority to define safety and the application of identical standards to US companies as prerequisites, making it a structure where finding common ground is difficult" [6]. This condition is fundamentally different from arms control negotiations during the nuclear era. Nuclear arms control was possible because both sides shared a common catastrophic risk of mutual assured destruction. In the AI domain, a common perception of threat at that level has not yet been formed.
3. Historical Precedents and Comparison of Similar Cases
On the surface, the establishment of the SI dialogue recalls the US-Soviet hotline or the nuclear arms control negotiation framework of the Cold War era. However, as the EAI analysis clearly points out, "this is merely a gesture of a crisis management nature and is difficult to view as a substantive arms control regime" [6]. Nuclear arms control involved verification systems, inspection mechanisms, and binding treaty language. In contrast, the SI dialogue does not contain concrete verification measures or binding norms, other than a scheduling agreement that "the next meeting would be held by November" [1]. To borrow CFR's phrasing, this summit is closer to an instance that exposed the "limits of photo-op summits" [5].
This is not the first time the Trump administration has attempted to change designations. The German public broadcaster Deutsche Welle highlighted precedents, noting that "the Gulf of Mexico was changed to the Gulf of America, and then Lake Ontario became Lake America," shedding light on the attempt to change AI to SI in the same context [18]. This demonstrates that the Trump administration has repeatedly followed a pattern of converting internationally accepted names or terms into objects of domestic political declarations. It is also similar to past cases in that the name change is not accompanied by substantive policy shifts. Just as renaming the Gulf of Mexico failed to alter international geographical naming practices, the term SI is not being voluntarily accepted by the international community, including China [1][15].
In terms of Big Tech regulation, this can be compared to the European Union's enforcement of GDPR in the late 2010s or antitrust lawsuits in the United States. Even then, regulatory investigations into giant platform companies proceeded along independent law enforcement procedures, separate from political rhetoric. Given that this FTC investigation was also initiated on the comprehensive grounds of "overall unfair practices violating the Federal Trade Commission Act" [4], it is closer to the routine exercise of authority by an existing regulatory agency rather than a specific political agenda. However, considering past precedents where antitrust investigations into Big Tech led to years of litigation, this investigation is also highly likely to be difficult to resolve in the short term.
4. Key Variables Shaping Future Developments
The first variable is the substantive outcome of the follow-up SI dialogue meeting scheduled for November in Shenzhen. An EAI analysis views that "whether notification procedures are concretized at the follow-up meeting in Shenzhen in November and the readjustment of export controls following the expiration of the Busan Agreement will serve as a test to gauge the practical credibility of this channel" [3]. If concrete notification procedures or verification measures are not agreed upon at this meeting, the SI dialogue is highly likely to solidify as a symbolic diplomatic device.
The second variable is the speed and scope of the FTC investigation's legal progress. Depending on whether the investigation leads to sanctions or corrective orders against specific companies, or merely stops at presenting guidelines for the industry as a whole, the investment and development strategies of the industry, including OpenAI and Anthropic, will vary. The possibility that the intensity of the investigation could be adjusted due to a change in the FTC chair or political pressure cannot be ruled out.
The third variable is whether China will accept the term SI. To date, China has never officially used this expression [1]. If China refuses to adopt it to the end, the difference in terminology between the US and China will remain a persistent point of friction symbolizing the difference in their AI governance philosophies. The framing of "Xi Jinping's human control or Trump's 'superintelligence'" chosen as a headline by the Nihon Keizai Shimbun [12] suggests that this rift stems from a fundamental difference in control philosophy rather than a simple translation issue.
The fourth variable is whether the regulatory and diplomatic tracks within the United States will integrate. As the EAI analysis points out, currently "the White House's cooperation track and the pressure tracks of Congress and regulatory agencies are operating separately" [6]. As long as this separation persists, agreements between leaders cannot constrain the execution of individual regulatory measures. From the perspective of third countries, including South Korea, this is precisely why dual monitoring is required to separately track the friendly signals from the White House and the regulatory measures of individual agencies such as the Department of Commerce and the FTC [6].
III. Recommended Responses
US FTC's AI Risk Investigation and the "AI-to-SI" Nomenclature Debate: Comprehensive Response Measures
1. Comprehensive Assessment and Recommended Responses
This issue is not a single event. It is the simultaneous overlap of three different tracks. The White House's SI renaming executive order [7], the agreement on the superintelligence dialogue at the US-China summit [1], and the FTC's investigation into OpenAI and Anthropic [4] each originated from independent institutional logics. As the EAI analysis points out, "the White House track, which signals cooperation, and the regulatory track, which continues investigations into advanced chip export controls and model distillation, are operating separately" [3]. What South Korea must guard against most when dealing with this issue is misreading these three tracks as a single, coherent US AI strategy.
The comprehensive assessment is as follows. The SI dialogue is a symbolic agreement. Immediately after the summit, President Trump reaffirmed his stance that the US would not share AI governance with China [6]. China has not officially adopted the term SI [1]. If this dialogue fails to concretize notification procedures at the follow-up meeting in Shenzhen in November, it is highly likely to converge toward a baseline scenario where only a nominal framework remains while substantive competition continues [9]. On the other hand, the FTC investigation is of a different nature. It is an investigation by an independent regulatory agency based on consumer protection jurisprudence, and it is an issue that can proceed regardless of election or diplomatic schedules [4][16]. The South Korean government and companies must design their response systems by separating these two.
The core of the recommended response is dual-track management. South Korea must distinguish between security-linked technology control domains and commercial AI cooperation domains, applying different risk assessment criteria to each. The point that policy priorities should not be linked to the symbolic announcements of the US-China dialogue has already been raised in EAI analyses [3][9]. At the same time, because trends in the FTC's safety investigations directly affect the US market entry strategies of South Korean AI companies, a separate, continuous tracking system is required.
2. Short-/Medium-/Long-Term Action Plans
The short-term (1–3 months) task is to establish a monitoring infrastructure. The Ministry of Foreign Affairs and the Ministry of Science and ICT should establish a channel to regularly review the progress and legal grounds of the FTC investigation. Given that the scope of the investigation is set to cover overall violations of the Federal Trade Commission Act [4], the possibility cannot be ruled out that the investigation's outcomes could indirectly affect the supply contracts or data processing methods of South Korean companies partnering with OpenAI or Anthropic. Simultaneously, South Korea must track the agenda and outcomes of the second US-China SI dialogue meeting scheduled for November to verify whether notification procedures are concretized [1][9].
The medium-term (6 months to 1 year) task is to refine the institutional response system. As the terminology change to SI has been officially reflected across the federal administration [7], there is a need to review the consistency of terminology in domestic laws and policy documents in preparation for the potential use of this term in future US government documents and draft regulations. However, as Danish experts point out, the prevailing view is that this concept is unlikely to take root internationally [15]. Therefore, there is little practical benefit for South Korea to preemptively transition its terminology. Instead, the key task during this period is to secure a voice in multilateral norm-setting tracks that are not subordinate to the bilateral US-China dynamic [6][9].
The long-term (1 year and beyond) task is to build independent supply chains and safety standard frameworks. If the structure in which the US regulatory track (FTC) and the White House track operate separately persists, South Korean companies should not hastily expand investments in China or contracts with US AI companies based solely on friendly signals from the US government [6]. It is advisable to establish a dual monitoring system that separately tracks changes in Department of Commerce regulations [6]. Furthermore, it is time to continuously track whether crisis management mechanisms, such as the mutual notification channel for AI accidents, are actually operational, and to review the necessity of introducing similar channels domestically [9].
3. Monitoring Indicators and Trigger Points
The first indicator is the outcome of the second US-China dialogue meeting in November. The key is whether notification procedures are codified into concrete protocols or remain limited to a principle-level reaffirmation [1][9]. The second indicator is the speed of the FTC investigation and whether sanctions are imposed. Whether Chairman Ferguson's investigation transitions to the official subpoena stage or is resolved through a settlement will serve as a test to gauge the future intensity of US AI regulation [4][16]. The third indicator is whether export controls are readjusted following the expiration of semiconductor export controls and the Busan Agreement [9]. This is a key signal showing whether the symbolic agreement of the SI dialogue actually translates into a relaxation of regulations against China.
Three trigger points can be established. The first is if China adopts the term 'SI' in official documents, which can be interpreted as a signal that symbolic rhetoric is transitioning into substantive diplomatic language. The second is if the FTC investigation results in specific corrective actions or fines against OpenAI or Anthropic, serving as a leading indicator of changes in regulatory intensity across the U.S. AI industry. The final trigger point is if the White House officially incorporates the SI framework into foreign policy documents (such as export controls or consultation agendas with allies). In this case, South Korea must separately review the issue of terminology usage in diplomatic channels.
4. Summary and Conclusion
At present, the assessment that the U.S.-China SI dialogue is a 'subordinate agenda item dependent on a trade truce package, such as tariff cuts worth approximately $30 billion,' remains persuasive [9]. Separately, the FTC investigation is an independent move originating from domestic consumer protection jurisprudence in the United States [4]. South Korea should perceive these two tracks separately and maintain a dual response system that distinguishes between security-linked and commercial domains [3][6][9]. While avoiding excessively linking policy priorities to symbolic diplomatic announcements, South Korea must simultaneously treat regulatory changes stemming from the FTC and adjustments to export controls by the U.S. Department of Commerce as separate subjects for constant monitoring. The follow-up meeting in Shenzhen in November and the expiration of the Busan Agreement are the most critical points of observation over the coming months.
References
[5] [Council on Foreign Relations (CFR)] The Limits of Trump and Xi’s Photo-Op Summit
[8] [Council on Foreign Relations (CFR)] How to Reach an AI Arms Control Deal With China
[11] [Council on Foreign Relations (CFR)] Why AI’s Biggest Rivals Are Suddenly Calling for Restraint
[12] [Nikkei Asia] Xi's human control or Trump's 'super intelligence': What's next for AI?
[14] [Exame] EUA e China criam canal de comunicação para incidentes com IA
[15] [Børsen] Trump vil ændre navnet på AI til SI: Eksperter er skeptiske
[17] [Dagens Industri] Myndighet utreder Anthropic och Open AI
[18] [DW (Deutsche Welle)] 'Super Intelligence': Is Trump right to rename AI?
[19] [Børsen] Myndighed indleder efterforskning af Open AI og Anthropic
[21] [DW (Deutsche Welle)] 'Super Intelligence': Is Trump right to rename AI?
[23] [Hürriyet Daily News] US, China agree to open AI incident reporting channel
*This text is an AI translation of an original written in Korean. Some translations or nuances may be inaccurate.
This report is an in-depth analysis planned by an EAI researcher, grounded in sophisticated AI-assisted research, and finalized by the EAI researcher.
How can I help you?