"‘Fable5 delisting’ sparks speculation of ‘weaponized’ AI regulation: OpenAI fears implication, Silicon Valley worries about crackdown on foreign AI talent"

Trump administration’s export control actions against Anthropic are triggering a chain reaction throughout the AI industry, pushing a long-suppressed sensitive issue in Silicon Valley to the forefront: Is the government “weaponizing” regulatory tools to limit AI companies’ reliance on foreign talent? A WallstreetCN article wrote that last Friday, U.S. Secretary of Commerce Howard Lutnick sent a letter to Anthropic, warning the company that providing access to its latest models, Fable and Mythos, to “foreigners”—including its own foreign employees—would require pre-approval via permits. This move effectively forced Anthropic to take down these models. On Monday, Anthropic held talks with officials of the Trump administration to discuss possible arrangements for re-launching the related models. On June 16, according to technology media outlet The Information, OpenAI’s Chief Strategy Officer Jason Kwon immediately stated in an internal Slack message that the company had “strongly” communicated to the government that building AI “requires top talent from around the world,” which is “one of the key reasons America maintains its lead in AI.” Additionally, it is worth noting that the impact of this incident extends beyond just Anthropic. According to The Information, during early discussions about the AI executive order signed by Trump this month, government officials proposed restricting foreign employees at AI companies from accessing frontier models, though this proposal did not make it into the final document. Meanwhile, some policy experts warn that the legal basis cited by the Commerce Department provides the government with a framework to establish a license regime for any AI model, which contradicts the White House’s previous opposition to mandatory licensing. This means a new, hard-to-quantify risk variable is emerging: unpredictable government policy. OpenAI officially responds, but deliberately keeps its distance from Anthropic OpenAI’s response to this incident is quite nuanced: on the one hand, the company clearly communicated its support for foreign talent to the government; on the other, its legal team drew a clear internal boundary. According to internal Slack messages obtained by The Information, OpenAI’s General Counsel Che Chang informed employees on Saturday not to “attempt to coordinate a response, or directly address the issue through contacts at Anthropic or other labs,” citing “antitrust rules” and advising the company “must proceed cautiously.” This stance reveals the complexity of the current situation: although major AI companies share common interests regarding foreign talent, the dual constraints of competition and compliance make it difficult for them to form a united front in publicly responding to the government. Foreign AI Talent: Silicon Valley’s Core Dependence and Policy Sensitive Area The high concentration of foreign researchers in the U.S. AI industry is the deep background of this controversy. According to reports, per MacroPolo’s estimates tracking global AI talent sources, in 2024, 38% of researchers who presented at top AI academic conferences completed undergraduate degrees in China, up from 29% five years ago; nearly three-quarters are now working in the U.S. Companies such as Anthropic, OpenAI, and Meta all recruit top AI researchers from overseas in large numbers. Aalok Mehta, Director of the Wadhwani AI Center at the Center for Strategic and International Studies, remarked: “Government concerns are particularly acute regarding Chinese nationals working on frontier AI models, especially ensuring that AI labs have robust control measures.” This concern is not new. In March, Emil Michael, Deputy Under Secretary for Research and Engineering at the U.S. Department of Defense, stated in a court declaration that Anthropic employs a large number of foreign nationals involved in model construction, which, given the requirement that employees comply with their respective national intelligence laws, “increases adversarial risk.” This statement appeared in a legal dispute between the Defense Department and Anthropic—Anthropic previously refused to allow the Pentagon to use its AI for large-scale domestic surveillance or fully autonomous weapons operations, after which the Defense Department labeled it as a supply chain risk. Analysts note that, aside from the uncertainty of massive data center investment and ambiguous demand outlook, unpredictable government policy is becoming another variable difficult to hedge against. It’s reported that the standoff between Anthropic and the government, alongside high vigilance from competitors like OpenAI, reveals: Even AI companies with relatively close government ties cannot fully avoid sudden policy shocks. If restrictions on foreign employees accessing frontier models expand further, it will directly impact the R&D capability and talent strategy of major AI labs, and in turn the competitive landscape across the industry. Legal Basis of the Commerce Department Letter: “A Significant Leap” The legal foundation of the action against Anthropic is also contentious. The Information reported that Lutnick’s letter cited Section 744.22(b) of the Export Administration Regulations, allowing the Bureau of Industry and Security to require export permits based on specific items posing risks of use by foreign military. The core claim in the letter is: Anthropic’s AI models pose “unacceptable risks” of being used for “foreign military intelligence purposes.” Notably, this export control applies to all foreigners globally, and also invokes “deemed export” rules, extending restrictions to foreign nationals residing inside the U.S. Joe Khawam, Executive Director of the Institute for Legal Reform and former State Department attorney, reviewed the letter’s wording and stated that the government’s use of this risk to cut off Anthropic’s model access for all foreigners is a “pretty big leap.” He added, “To achieve the outcome they wanted—getting Anthropic to take down the models—I think they had to do it.” Adam Thierer, Senior Fellow at the conservative R Street Institute, described the move as entering an “era of provisional, indirect, backdoor AI licensing,” warning, “Once you do this, it’s not easy to put the genie back in the bottle.” Earlier WallstreetCN articles noted that a direct trigger for this export control was an Amazon report revealing security vulnerabilities in Anthropic’s latest models. Reportedly, former AI and crypto executive David Sacks stated on social platform X that the government acted “reluctantly” because Anthropic was unwilling to cooperate on the issue. Anthropic responded in a blog post that the vulnerabilities were “simple” and commonly found in other publicly available models. Risk Warning and Disclaimer The market carries risks, and investment should be cautious. This article does not constitute personal investment advice and does not consider individual users’ special investment goals, financial conditions, or needs. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific situation. Investments made accordingly are at your own risk.