Anthropic’s Downfall for Upholding Ethics
Flagship Models Hit by Sweeping Export Controls
OpenAI Secures Monopoly by Yielding to Authority
White House’s Blueprint for Technological Supremacy

U.S. Monopoly System Now Weaponized

Today, the rapidly expanding generative artificial intelligence (AI) market has evolved into a ‘global geopolitical battlefield,’ where national computing infrastructure, astronomical capital, and highly sophisticated political influence intersect and compete.


In the U.S. AI ecosystem, the conditions for winning are clear. The company that can secure the world’s most advanced proprietary large language models (LLMs) and large multimodal models (LMMs), while also winning security and military contracts backed by the massive budgets of the Department of Defense (DoD) and other U.S. federal agencies, will emerge victorious. This company, leveraging its position, will gain absolute hegemony over the market.


[Jung Inkyo Column] U.S. Anthropic Regulation and Hegemony... National Survival Hinges on 'Sovereign AI' View original image

Amid this fierce struggle for hegemony, the global AI industry and the U.S. political establishment are now experiencing unprecedented, sharp conflict between two values: ‘national security and unlimited innovation’ versus ‘technological ethics and controllability.’ Notably, OpenAI and Anthropic have emerged as the central players, standing at the very frontlines of this conflict. Anthropic was founded by former OpenAI researchers who favored a safety-first approach.


The confrontation began earlier this year. At that time, Anthropic was progressing smoothly, having signed a major contract worth approximately 200 million dollars (about 300 billion won) to supply large-scale AI models and consulting to the U.S. federal government and Department of Defense. However, Anthropic took a strong stance, declaring it would absolutely not allow its cutting-edge AI models—‘Claude’ and the ‘Mythos’ series—to be directly used for ‘mass surveillance’ or in the development and targeting processes of ‘lethal autonomous weapons systems’ (LAWS) within the U.S. military.


In contrast, the Trump administration and top Pentagon officials, prioritizing overwhelming military power and technological supremacy, were adamant in their demands. The Trump administration insisted on an ‘unrestricted right to use the technology,’ with no constraints or filtering on AI models. When Anthropic ultimately refused the government’s demands and maintained its ethical code, the Department of Defense immediately took harsh retaliatory action. In March, the Pentagon designated Anthropic as a ‘supply chain risk’ company that could threaten national security, added it to the official blacklist, and abruptly canceled the existing 200 million dollar contract. As a result, Anthropic, by upholding its ethics and values, became a loser in the federal government contract market.


Anthropic logo. Reuters·Yonhap News

Anthropic logo. Reuters·Yonhap News

View original image

By contrast, the absolute market leader, OpenAI, chose a thoroughly pragmatic path. OpenAI willingly accepted the Pentagon’s strong pressure and demands, allowing its generative AI models to be used ‘for all lawful purposes’ across the board by the military. While there were some stated conditions, in reality this amounted to a strategic capitulation, effectively granting the government and military control over its AI. In return, OpenAI won a massive contract. On June 1, the Department of Defense awarded a huge confidential AI service project to a consortium of eight tech companies led by OpenAI. Meanwhile, Anthropic’s name was completely absent from the list for this major national project.


The government’s all-out pressure on Anthropic did not stop with the cancellation of the defense contract—it escalated further. On June 12, the U.S. Department of Commerce, citing national security, imposed powerful export controls to prevent foreign nationals from accessing Anthropic’s latest flagship models, ‘Fable 5’ and ‘Mythos 5.’ Anthropic strongly objected, arguing that the technical risks (‘jailbreaking’) associated with Mythos 5 are also issues for its competitor OpenAI, and that singling out only Anthropic’s models for global-level service restrictions constitutes an egregiously unfair ‘targeted sanction.’ Nonetheless, these protests were ignored.


These export controls by the Commerce Department are now tightening the noose around Anthropic’s neck just ahead of its highly anticipated IPO in October. For Anthropic, the economic blow is devastating. While the company’s management has tried to reassure the market through official statements suggesting a legal settlement is possible, in reality, market expectations remain extremely bleak.


Behind all this lies the firm and emotionally charged political hostility of U.S. President Donald Trump. President Trump has fiercely criticized Anthropic as a ‘radical left corporation that prioritizes its own hypocritical moralism over America’s national security’ and ‘woke zealots hiding behind a wall of technology.’ The administration appears poised to use these export controls as leverage, tightening the regulatory grip until Anthropic completely surrenders and submits 100% to the government’s military and security guidelines.


The Anthropic situation is shaking the very foundations of the AI ethics system that the international community has painstakingly built over the past decade. Moreover, the sweeping measures imposed by the Pentagon and Commerce Department on Anthropic are sending a profoundly disruptive message to the global technology market. Of even greater significance, this incident will go down in the history of global trade as the world’s first actual case of high-performance AI regulation and export control in the form of a service model (SaaS). Until now, Korea and other major countries have primarily monitored U.S. export controls on hardware items such as semiconductor equipment, advanced physical components, and raw materials, or on traditional source code technologies, in order to manage supply chain risk. Now, however, the world is confronted with a new risk scenario: the U.S. government can, based on political and geopolitical judgment, block API access to AI models residing on the servers of major American tech companies overnight.


There is only one answer for survival amid a U.S.-centered AI monopoly regime and the ever-present threat of asymmetric technological control being weaponized at any moment. It is urgent to establish an independent AI security consultative body that is immune to foreign political winds. In addition, building and advancing a ‘sovereign AI and independent infrastructure ecosystem’ must be pursued as a matter of national survival to cope with AI model risk.



[Jung Inkyo Column] U.S. Anthropic Regulation and Hegemony... National Survival Hinges on 'Sovereign AI' View original image

Jung In-gyo, Professor of International Trade at Inha University (former Chief Negotiator for Trade)


This content was produced with the assistance of AI translation services.

© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.

Today’s Briefing