President Donald Trump has reversed his position on Anthropic PBC being a national security threat, stating he now views the company as having 'behaved very responsibly' following the Commerce Department's recent restrictions. In a recent interview with Axios, Trump acknowledged that his concerns had changed dramatically, saying 'Well, not now, but a week ago, maybe' when asked whether he viewed Anthropic as a security risk. The president revealed that Amazon, both a competitor and part-owner of Anthropic, alerted the administration about serious vulnerabilities in the company's AI models. Trump explained that 'It was a competitor and a part owner that turned Anthropic in. They didn't like what they were doing. They were very concerned. Think of it, it's a part owner, and I think it worked out very well.' He added that 'He responded to us very quickly because you know it's a tremendous liability. People get put in prison immediately for that. You can't play games with that.'
The US has ordered Anthropic PBC to disable access to its most advanced AI platforms for all foreign nationals after discovering it's possible to 'jailbreak' or bypass the guardrails of the new Fable 5 AI model. According to reports from Business Standard, this development has prompted world leaders including Canadian Prime Minister Mark Carney to warn of the risks of depending on just a handful of AI tools. The Commerce Department's directive issued on June 12 marked one of the most significant government interventions into the operations of an artificial intelligence company. The move has created fresh challenges for Anthropic only weeks after the company confidentially filed for an initial public offering, with its latest valuation reportedly surpassing ₹7,500 crore ($900 billion). The restrictions prevent foreign governments, companies and individuals from accessing Anthropic's newest and most advanced AI systems, with the company forced to disable access because the scope of the order affected a broad range of users outside the United States. Washington cited concerns over so-called 'jailbreaking', where prompts are designed to bypass built-in safety restrictions, arguing that such vulnerabilities could create national security risks.
The White House and Anthropic are now drafting a joint risk framework to assess AI security vulnerabilities and guide government intervention across future incidents involving frontier AI models. According to Politico, the shared standard will evaluate the severity of AI security flaws and determine when the government should intervene. The framework follows the export controls imposed over a so-called jailbreak in Fable 5 and Mythos 5, moving from open confrontation to direct technical collaboration. Negotiators are defining how far safeguards were bypassed, what capabilities were exposed, and the real-world consequences of any breach. The framework could serve as a template for all future interactions between governments and AI developers across the industry, potentially providing a White House mechanism for evaluating future AI risks without resorting to emergency interventions each time a vulnerability is discovered.
The European Commission has responded to the US restrictions by emphasizing the need for technological independence while calling for international cooperation on frontier AI systems. According to MLex, Henna Virkkunen, the European Commission's digital chief, has stated that Europe is an economic opportunity, not a security risk. The Commission is reviewing the practical impact of the restrictions on European users and businesses that rely on advanced AI tools, with officials warning that critical AI services controlled by foreign companies or governments may not always remain available to international users. These comments reflect growing concerns among policymakers that access to key technologies may increasingly become tied to geopolitical decisions, prompting calls for more coordinated international approaches to AI development. At the Group of Seven summit in Évian, Indian Prime Minister Narendra Modi called for broader and more inclusive access to American artificial intelligence models, adding another dimension to the growing international debate over access, regulation and security.
The US restrictions on Anthropic have sparked a global movement toward AI sovereignty, with countries worldwide reassessing their reliance on foreign AI technologies. According to Business Standard, the incident has prompted world leaders including Canadian Prime Minister Mark Carney to warn of the risks of depending on just a handful of AI tools. The development has fueled calls for diversification and self-reliance in AI development across multiple nations, as countries recognize the strategic importance of controlling their own AI infrastructure and capabilities. The restrictions have particularly highlighted concerns about technological dependence, with officials warning that access can be limited or suspended with little warning. The latest dispute adds to months of tensions between Anthropic and the US government, with the company having been engaged in a prolonged disagreement with the Pentagon over additional safeguards for military applications of its artificial intelligence tools.
Sung Kim, CEO of Upstage, emphasized the strategic importance of AI independence during a media briefing in Seoul. According to Business Standard, Kim stated that 'AI is no longer just a service or a tool we use; it has become a strategic national asset'. He noted that countries controlling foundational AI technologies, such as the US and China, can curtail access at any time, making self-reliance crucial. The company recently acquired web portal Daum from Kakao Corp. as part of its business strategy to combine AI models, AI agents and consumer portal into a single platform aimed at bringing AI to the general public. The US government has said it expects the controls to be temporary and hopes Anthropic can address the safety concerns that led to the decision. Anthropic has developed a reputation as one of the most cautious companies in the race to develop advanced artificial intelligence systems, following a safety-focused strategy known as Constitutional AI, in which its models are trained to evaluate and critique their own outputs against a set of written principles.
Experts in South Korea's science and technology community view the US restrictions as a direct warning for Korea's AI development strategy. Choi Ki-young, former Minister of the Ministry of Science and ICT, noted that the incident will further heighten awareness that AI could be treated as an important national strategic asset. He emphasized that high-performance AI models can be used for both cyberattacks and defense, making it a major threat if one country monopolizes such capabilities. Kim Nam-kuk, a physician-scientist at Ulsan University College of Medicine, predicted the move will stoke global AI development competition and potentially weaken the US-centric global AI ecosystem, leading to a reorganization into a new tripolar system. Lee Sang-yeop, distinguished professor at KAIST, stressed that Korea should prioritize securing AI capabilities that operate without overseas access in key areas like defense, healthcare, research, and administration, rather than immediately catching up with U.S.-level ultra-large models. The relationship between the White House and Anthropic appears to be on the mend, though the technical work of setting AI safety standards and defining international cooperation remains uncertain over the coming months.
