WASHINGTON, D.C. — June 15, 2026 — The U.S. Department of Commerce on Friday ordered artificial-intelligence company Anthropic to restrict access to its two most powerful systems, Fable 5 and Mythos 5, significantly limiting international use of the models and marking one of the most aggressive federal interventions yet in the rapidly evolving artificial intelligence industry.
According to Anthropic, the export-control directive was delivered by letter at 5:21 p.m. ET Friday and originated from Commerce Secretary Howard Lutnick and the department’s Bureau of Industry and Security. The action followed warnings from Amazon Chief Executive Officer Andy Jassy, who reportedly alerted senior administration officials that internal testing had revealed potential security vulnerabilities in the models.
The dispute began after Amazon researchers conducted a series of tests designed to probe the systems’ safeguards. According to accounts of the matter, the researchers were able to use carefully crafted prompts to bypass certain protections and generate information that could potentially assist in cyberattacks — material the systems were designed to block.
Jassy reportedly escalated those findings to senior officials in Washington, setting off a series of discussions inside the administration regarding whether the models presented a national-security concern.
Government researchers subsequently conducted their own evaluations of the systems. Officials then reportedly presented Anthropic with a choice: address the identified vulnerabilities immediately or face restrictions on deployment of the affected models.
According to a senior administration official, President Donald Trump ultimately approved the action while expressing concern that excessive regulation could slow American innovation in artificial intelligence.
The resulting order was unusually broad.
Rather than limiting access only overseas, the directive reportedly prohibited use of Fable 5 and Mythos 5 by foreign nationals regardless of location, including individuals located inside the United States. Anthropic stated that it did not have a practical method to selectively block only foreign users and therefore suspended access to the two models more broadly while complying with the order.
The company said access to its other AI products remains available.
Anthropic has publicly complied with the directive while strongly disputing the government’s conclusions.
The company characterized the issue as a narrow jailbreak scenario and argued that the vulnerabilities identified by Amazon were limited in scope and already understood within the industry. Anthropic warned that if the same standard were applied universally, it could substantially hinder development and deployment of advanced AI systems across the sector.
The company further noted that it had implemented extensive safeguards designed specifically to prevent cybersecurity misuse and argued that no AI system is entirely immune from determined attempts to circumvent protections.
The dispute places Amazon in an unusual position.
The technology giant is both one of Anthropic’s largest investors and a major provider of cloud-computing infrastructure used to train and operate Anthropic’s models. By bringing the concerns to federal officials, Amazon effectively placed national-security considerations ahead of a business relationship involving billions of dollars in investment and infrastructure commitments.
For Anthropic, the impact was immediate.
The company said two of its flagship AI systems, which collectively reach hundreds of millions of users worldwide, were effectively removed from broad international availability pending further review.
The broader significance may extend far beyond a single company.
The United States has previously restricted exports of advanced semiconductors and computing hardware used to train artificial intelligence systems. However, industry observers note that this appears to be among the first major instances in which federal authorities directly restricted access to an AI model itself rather than the hardware powering it.
The action could establish a new precedent for government oversight of advanced AI systems and may signal the emergence of a de facto approval framework under which regulators determine when certain models can be deployed internationally.
Such a framework would represent a significant shift from the administration’s broader approach toward artificial intelligence, which has generally emphasized voluntary cooperation and innovation rather than formal licensing requirements.
Investors are closely watching the implications for both AI developers and the companies supporting them.
Because Anthropic remains privately held, the immediate public-market impact is most visible through Amazon (NASDAQ: AMZN), which closed Friday at $238.55, down 1.23%. The decline occurred before the directive was reportedly issued and was largely attributed to broader concerns surrounding artificial-intelligence spending and regulation rather than the specific action against Anthropic.
Administration officials have indicated the restrictions may be temporary and could be lifted if Anthropic satisfies federal security concerns following additional review.
For now, the episode raises a fundamental question facing the artificial-intelligence industry: who ultimately decides when a powerful AI system is safe enough to remain widely available — the company that develops it, or the government that has the authority to restrict access.
Anthropic maintains that the government’s action is based on a misunderstanding of the risks involved and says it is actively working with federal officials in hopes of restoring broader access to the models.
JBizNews Desk — Technology
© JBizNews.com All Rights Reserved. Reproduction or distribution without written permission is prohibited.
