Original Title: Inside the whirlwind 24 hours that led the White House to slap export controls on Anthropic
Original Author: Sophia Cai and Cheyenne Haslett, Politico
Original Translator: Peggy
Editor's Note: Anthropic had just released the Fable 5 model a few days prior when it was suddenly forced offline. The reason was that the Trump administration, citing national security, demanded restrictions on access to the advanced AI models Fable 5 and Mythos 5 by foreign nationals; due to the difficulty in accurately distinguishing user identity and access permission, Anthropic ultimately chose to disable access to the relevant models for all clients.
Behind this unexpected withdrawal was a high-pressure game played out between the White House and Anthropic within 24 hours. Concerning whether the safety barriers around Fable 5 could be circumvented, Anthropic CEO Dario Amodei engaged in multiple tense phone calls with senior government officials, including those from the Treasury Department, the Department of Commerce, and the White House's cyber chief. The government believed the model could be used to identify software vulnerabilities, thereby posing a national security risk; Anthropic emphasized that the issues in question did not equate to "universal jailbreaking" and criticized the government actions as lacking transparency, clarity, and legally backed procedures based on technical facts.
The crux of the matter was not just a model being taken offline, nor was it simply a breakdown in communication between the White House and an AI company, but that advanced AI models are being redefined as a potential national security asset. When model capabilities enter sensitive scenarios such as cybersecurity, vulnerability exploitation, and intelligence analysis, the government is no longer satisfied with corporate self-assessments and voluntary reviews but begins to employ more stringent policy tools like export controls to directly intervene in the rhythm of model releases.
Ironically, Anthropic has long been one of the most vocal companies advocating for regulation in the AI industry, only to find itself as the target of regulatory action. The government believed its response to security vulnerabilities was not serious enough, while Anthropic viewed the White House's reaction as an overreach. Behind the disputes between the two parties lies a new question that the AI industry must confront in the future: Who has the authority to determine whether a model is safe enough? Is a company's self-certification of safety sufficient? How will rapid interventions under the guise of national security alter the compliance costs, release rhythm, and global competitive landscape for AI companies?
The Anthropic incident could mark a watershed moment in the regulation of advanced AI. In the future, the release of cutting-edge models will no longer be just a product launch, but more like a high-risk deployment involving national security, geopolitical considerations, capital markets, and industrial order.
Below is the original text:
Before the Trump administration imposed comprehensive export controls on Anthropic, senior White House officials attempted to persuade the AI company to voluntarily withdraw a newly released model in an emergency negotiation that lasted 24 hours. Multiple government officials believed that the model posed a national security risk.
According to two government officials and a senior White House official, during this turmoil, Anthropic CEO Dario Amodei had several tense phone calls with multiple high-ranking government officials, including U.S. Treasury Secretary Scott Bessent and White House Cyber Chief Sean Cairncross. Due to not being authorized to publicly discuss the matter, those involved requested anonymity.
This incident has brought to the forefront an increasingly urgent question: How should the White House draw the line between innovation and safety faced with rapidly iterating advanced AI models with expanding capabilities?
These details of the calls had never been reported before.
Night Before the Withdrawal: Safety Controversy
After the government implemented export controls, Anthropic was forced to take its new model Fable offline. The model had been publicly released for only a few days. Anthropic had previously assured the public that the model was safe and controllable, but shortly after its release, senior government officials began to reevaluate whether its safety barriers were as robust as the company claimed.
Two government officials and a senior White House official stated that on Thursday, just two days after the public release of Fable, Amazon CEO Andy Jassy expressed concerns to the White House that the model's safety barriers might be at risk of being circumvented.
A person familiar with Amazon's communications indicated that Amazon was providing feedback in response to government requests regarding potential risks.
By Friday morning, the issue had escalated to the highest levels of the White House.
According to government officials and a senior White House official, Bessent, Cairncross, White House Chief of Staff Susie Wiles, and other senior officials held a meeting to discuss the model and government response. One knowledgeable individual noted that Bessent was remote attending the meeting because he was en route to Houston for a scheduled public event.
After the meeting, the government attempted to contact Amodei, but was told he was temporarily unavailable due to a health retreat. Both a government official and a senior White House official stated this.
An Anthropic spokesperson denied this claim, stating, "This is completely untrue."
A person close to Anthropic indicated that the government first requested to contact Amodei around noon, and Amodei was on a call with senior officials within 1 hour and 15 minutes. This individual noted that during the time Amodei was unavailable, Anthropic had proposed other executives to communicate on his behalf.
Once the government finally reached Amodei, he participated in three calls. A senior White House official and a government official stated that about half a dozen high-ranking officials, including Cairncross, Bessent, and Secretary of Commerce Howard Lutnick, participated in the calls.
A senior White House official noted that other White House staff and government officials also participated in parts of the calls, including Deputy Secretary of Commerce for Industrial and Security Affairs Jeffrey Kessler, White House Staff Secretary Will Scharf, Deputy Chief of Staff Richard Walters, and presidential policy advisor Walker Barrett.
During the calls, Amodei tried to clarify the misunderstandings he believed existed. He rebutted the government's security concerns, defended Anthropic's protective mechanisms, and emphasized that the circumvention methods that had emerged were specific issues under certain scenarios and did not equate to a broader meaning of "jailbreaking"—which would mean completely bypassing the safety barriers set by Anthropic.
In a blog post published after the export controls were implemented, Anthropic stated, "Currently, no tester has been able to find a universal jailbreak method—a way to broadly bypass model safety measures and unlock significant network capabilities." The company also stated that completely eliminating all jailbreak attempts was not realistic for Anthropic or any other AI company.
Anthropic also defended its security system, stating that its protective mechanisms were "so strong that many users complain they are too broad."
However, Cairncross and Bessent were not persuaded by Amodei's explanations. A White House official stated that Amazon's findings were submitted to the National Security Agency for review, which deemed the relevant materials sufficient to constitute "evidence."
According to senior White House officials and two government officials, the government urged Anthropic to proactively take down the model and cooperate with the government to fix the vulnerabilities. Amodei requested more time and information but did not promise to take down the model. A senior White House official stated that during one of the calls, Bessent directly told Amodei that he was making a "bad decision."
Regulatory Iron Fist Falls: Fable 5 and Mythos 5 Encounter Controls
Shortly after the calls ended, the Trump administration imposed export controls on the Fable 5 and Mythos 5 models. Anthropic stated that the government invoked national security authority to prohibit foreign nationals from using these models. The company indicated that the "practical effect" of the order was that, to ensure compliance, Anthropic had no choice but to "suddenly disable" access to the relevant models for all clients.
A senior White House official stated, "We pleaded with them to cooperate with us for several hours; export controls were a last resort. This is not something we wanted to do, but we had no choice."
After the report was released, a source close to Anthropic argued that the company had not been given the so-called "voluntary cooperation" option.
The individual stated, "The White House only gave a 90-minute request to take down the model, with no details about the actual threat. There was never any plea or request for them to cooperate; there was only an announced 90-minute deadline."
A senior White House official stated that White House officials had heard Amodei compare the dangers of Anthropic's technology to a nuclear bomb. Therefore, when this CEO, who has been advocating for AI safety regulation, expressed unwillingness to take down the system to fix known security vulnerabilities, government officials found it difficult to understand.
Anthropic has long been regarded as one of the most vocal companies advocating for regulation within the AI industry. The company has repeatedly argued that regulation frameworks should be established to address the global security risks and employment impacts that could arise from the rapid development of AI.
Three individuals familiar with the government's mindset indicated that Amazon was not the only company that raised concerns with the government.
One individual stated, "The core issue is that Anthropic lacked a sufficiently serious attitude toward this matter. If Anthropic took it seriously and did not downplay it as an isolated incident, and instead proactively fixed or paused access, this would not have escalated to this point."
Another person close to Anthropic countered that the so-called "jailbreaking" did not imply a failure of Fable 5's security system, and pointed out that the company maintained communication with the government before releasing Fable. This individual noted that in multiple rounds of communication, the government did not oppose the release of Fable.
After the government implemented export controls, Anthropic stated in a blog post that it would comply with government directives but also considered the action to be an overreaction.
Anthropic stated, "As we have previously stated openly, we believe the government should have the authority to prevent unsafe deployments, but this should be based on a transparent, fair, clear, and technically factual legal process. This action does not align with these principles."
A White House official authorized to speak anonymously stated that innovation remains the "top priority" for the White House, but "we must also prioritize safety."
Amazon declined to disclose specific details regarding its communications with the government in its statement. An Amazon spokesperson noted, "It is not uncommon for the government to consult us about potential security risks. When such situations occur, we do not disclose the details of related discussions."
In early April this year, Anthropic announced that its latest powerful model Mythos would be open only to a limited number of tech companies and cybersecurity firms to help them test software vulnerabilities. Anthropic stated at the time that due to the model's excessive capabilities, if it fell into the hands of improper users, it could lead to severe consequences, hence the need to limit its release scope.
The introduction of Mythos sparked a series of meetings between Amodei and senior White House officials. Both parties described these meetings as productive. Subsequently, multiple rounds of discussions were held regarding how to regulate advanced AI models, ultimately leading to a recent executive order requiring companies to voluntarily submit models for government review before broad deployment.
This week’s publicly released Fable 5 was described by Anthropic as a "Mythos-level model," but equipped with additional safety measures to make it available to general users. The model had been reviewed by both the U.S. government and the UK AI Safety Institute.
However, after alleged safety flaws were disclosed, multiple government officials believed that the model must be taken offline immediately.
On Saturday morning, former White House AI chief and staunch opponent of regulation David Sacks posted on X supporting the government’s export controls on Anthropic.
Sacks stated that he did not consider the "jailbreaking" to be simple or insignificant, nor did he view export controls as the government attempting to gain broader control over the AI industry. Sacks wrote, "The government now wants Anthropic to fix the safety issues, then lift the export controls, and bring Fable back to the public release state. The government hopes all this happens quickly. Frankly, the government feels confused that Anthropic, which has repeatedly stated that safety requests are its top priority, is now unwilling to cooperate."
Sacks and other government officials have previously criticized Anthropic multiple times for perceived left-wing political bias and have accused the company of fostering panic through its advocacy for stronger industry regulations and warnings of large-scale employment impacts.
Earlier this year, Defense Secretary Pete Hegseth and the Pentagon heightened the existing divide between the government and Anthropic. On March 3, due to Anthropic's refusal to allow its AI tools to be used for mass domestic surveillance and autonomous weapon systems, the Pentagon classified the company as a supply chain risk.
On Saturday, Sacks stated that the existing disputes between the government and Anthropic were separate from the decision to impose export controls. Sacks wrote, "The government values Anthropic's technological capabilities and believes this issue, although serious, should be easy to resolve. The ball is now in Anthropic’s court."
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
