Fable5이 강제로 오프라인되기 24시간 전, 백악관에서는 무슨 일이 일어났을까?규제를 가장 지지하는 AI 기업, 오히려 규제의 철퇴를 맞다

• 핵심 관점: 미국 정부는 Anthropic의 신규 모델 Fable 5의 안전장치가 우회될 가능성을 우려하여 국가 안보 위협으로 간주, 24시간에 걸친 치열한 협상 끝에 드물게 수출 통제권을 발동해 모델을 강제로 내리게 했다. 이 사건은 첨단 AI 모델이 잠재적인 국가 안보 자산으로 간주되기 시작했으며, 규제 방식이 기업의 자율 규제에서 정부의 강제 개입으로 전환되고 있음을 의미한다.
• 핵심 요소:
  1. 정부는 해당 모델이 소프트웨어 취약점을 식별하는 데 악용될 수 있어 국가 안보 위협이 된다고 판단한 반면, Anthropic은 관련 문제가 ‘범용적 제프 돌파(탈옥)’가 아니라고 주장했다.
  2. 이번 사건은 아마존이 백악관에 우려를 표명하면서 시작되었으며, 이후 미국 재무부, 상무부 및 백악관 사이버 보안 책임자 등 고위 관계자들이 Anthropic CEO와 여러 차례 긴장된 통화를 진행했다.
  3. 정부는 Anthropic에 자발적인 모델 내리기를 요구했으나 받아들여지지 않자, 국가 안보 권한을 발동하여 Fable 5 및 Mythos 5에 대한 수출 통제를 시행했고, 이로 인해 모든 고객에 대한 모델 사용이 금지되었다.
  4. 규제를 적극적으로 촉구해 온 AI 기업인 Anthropic이 오히려 규제의 철퇴를 맞은 셈이 되었으며, 회사 측은 정부의 조치가 투명성과 기술적 사실에 기반한 법적 절차를 결여했다고 비판했다.
  5. 이번 사건은 분수령이 될 것이며, 향후 최첨단 모델의 출시는 단순한 제품 출시가 아니라 국가 안보와 지정학적 위험을 수반하는 고위험 배포 과정에 가까워질 것이다.

Original headline: Inside the whirlwind 24 hours that led the White House to slap export controls on Anthropic

Original authors: Sophia Cai and Cheyenne Haslett, Politico

Original translation: Peggy

Editor's note: The Fable 5 model, released just days ago by Anthropic, was suddenly taken offline. The trigger was the Trump administration's demand, citing national security concerns, to restrict foreign nationals' access to Anthropic's advanced AI models, Fable 5 and Mythos 5. Unable to precisely distinguish user identity and access permissions, Anthropic ultimately chose to disable the relevant models for all customers.

Behind this sudden takedown was a high-pressure 24-hour standoff between the White House and Anthropic. Anthropic CEO Dario Amodei engaged in multiple tense phone calls with senior government officials, including the Treasury Secretary, Commerce Secretary, and the White House Cyber Director, over whether Fable 5's safety guardrails could be bypassed. The government believed the model could be used to identify software vulnerabilities, posing a national security risk. Anthropic, on the other hand, emphasized that the issue was not a "general jailbreak" and criticized the government's actions as lacking transparent, clear, and technically-based legal procedures.

The crux of this incident extends beyond a single model takedown or a communication breakdown between the White House and an AI company. It signifies that advanced AI models are being redefined as potential national security assets. When model capabilities enter sensitive domains like cybersecurity, vulnerability exploitation, and intelligence analysis, governments are no longer satisfied with corporate self-assessments and voluntary reviews. They are beginning to employ harder policy tools, such as export controls, to directly intervene in model release timelines.

Ironic as it may be, Anthropic has been among the most vocal advocates for regulation within the AI industry, yet it has become the target of the regulatory hammer. The government believes its response to the security flaw was not serious enough, while Anthropic argues the White House overreacted. Underlying this dispute is a new question the AI industry must face: Who has the authority to determine if a model is sufficiently safe? Is a company's self-certification of safety enough? And how will rapid interventions under the guise of national security alter compliance costs, release cycles, and the global competitive landscape for AI companies?

The Anthropic incident could become a watershed moment in advanced AI regulation. In the future, the release of cutting-edge models will no longer be just a product launch but a high-risk deployment involving national security, geopolitics, capital markets, and industrial order.

Below is the original text:

Before the Trump administration imposed sweeping export controls on Anthropic, senior White House officials engaged in a 24-hour emergency mediation effort, trying to persuade the AI company to voluntarily take down a newly released model that multiple government officials believed posed a national security risk.

According to two government officials and one senior White House official who spoke on condition of anonymity because they were not authorized to discuss the matter publicly, Anthropic CEO Dario Amodei held several tense phone calls with high-ranking officials during this episode, including Treasury Secretary Scott Bessent and White House Cyber Director Sean Cairncross.

This incident also brings a pressing question to the forefront: How should the White House draw the line between innovation and security when faced with advanced AI models that iterate rapidly and whose capabilities constantly expand beyond current boundaries?

These details of the calls have not been previously reported.

The Night Before the Takedown: The Safety Controversy

After the government imposed export controls, Anthropic was forced to take down its new model, Fable, which had been publicly available for only a few days. Anthropic had previously assured the public of the model's safety and controllability, but shortly after its release, senior government officials began reassessing whether its safety guardrails were as robust as the company had claimed.

Two government officials and one senior White House official stated that on Thursday, two days after Fable's public release, Amazon CEO Andy Jassy expressed concerns to the White House that the model's safety guardrails could potentially be bypassed.

A person familiar with Amazon's communications related to the matter said Amazon was responding to a government request for feedback on potential risks.

By Friday morning, the issue had escalated to the highest levels of the White House.

According to the government officials and the senior White House official, Bessent, Cairncross, White House Chief of Staff Susie Wiles, and other senior officials convened to discuss the model and the government's response. One of the informed sources noted that Bessent was en route to Houston for a scheduled public event and participated remotely.

Following the meeting, the government attempted to contact Amodei but was told he was temporarily unavailable due to a health retreat, according to one government official and one senior White House official.

An Anthropic spokesperson denied this account, stating, "That is completely false."

A person close to Anthropic said the government first requested to contact Amodei around noon, and Amodei was on the phone with senior officials within 1 hour and 15 minutes. This person stated that during the brief period Amodei was unavailable, Anthropic offered to have other executives communicate on his behalf.

Once the government finally reached Amodei, he participated in three separate calls. The senior White House official and one government official said about half a dozen senior government figures were involved in the calls, including Cairncross, Bessent, and Commerce Secretary Howard Lutnick.

The senior White House official added that other White House staff and government officials, including Commerce Under Secretary for Industry and Security Jeffrey Kessler, White House Cabinet Secretary Will Scharf, Deputy Chief of Staff Richard Walters, and Presidential Policy Advisor Walker Barrett, also participated in some of the calls.

During the calls, Amodei attempted to clarify what he believed were misunderstandings. He pushed back against the government's safety concerns, defended Anthropic's safeguards, and emphasized that the bypass method that had been demonstrated was a specific contextual issue, not equivalent to a broader "jailbreak"—where a model is completely freed from Anthropic's safety guardrails.

In a blog post published after the export controls were imposed, Anthropic stated: "Thus far, no testers have found a general jailbreak—a jailbreak that would enable broad removal of the model's safety measures and unlock a large set of cyber capabilities." The company also said that it is currently unrealistic for Anthropic or any other AI company to completely eliminate all jailbreak attempts.

Anthropic also defended its safety system, stating its guardrails are "so strong that many users complain about them being overly broad."

However, Cairncross and Bessent were not convinced by Amodei's explanations. A White House official said Amazon's findings had been submitted for review by the National Security Agency (NSA), which considered the materials sufficient to constitute "evidence."

According to the senior White House official and two government officials, the government urged Anthropic to voluntarily take down the model and coordinate with the government to fix the vulnerability. Amodei requested more time and information but did not commit to taking the model down. The senior White House official said that in one of the calls, Bessent directly told Amodei that he was making a "bad decision."

The Regulatory Hammer Falls: Fable 5 and Mythos 5 Hit by Controls

Shortly after the calls concluded, the Trump administration imposed export controls on both 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 said the "practical effect" of this order forced Anthropic to "abruptly disable" access to these models for all customers to ensure compliance.

A senior White House official stated: "We begged them to work with us for hours. Export controls were a last resort. It's not what we wanted to do, but we felt we had no other choice."

Following this report, a person close to Anthropic countered, arguing that the company was not given the option of so-called "voluntary cooperation."

This person said: "The White House gave us 90 minutes to take the model down, with no details about the actual threat. There was never any begging or asking to work with them. There was just a declared 90-minute ultimatum."

A senior White House official said White House officials had heard Amodei compare the danger of Anthropic's technology to a nuclear bomb. Therefore, government officials found it difficult to understand why a CEO who had been a proponent of AI safety regulation would be unwilling to take down a system to fix known security vulnerabilities.

Anthropic has long been seen as one of the most vocal advocates for regulation within the AI industry. The company has repeatedly argued that regulatory frameworks are needed to address the global safety risks and job displacement potentially caused by the rapid development of AI.

Three people familiar with the government's thinking said Amazon was not the only company that raised concerns with the government.

One of them said: "The core of the issue is that Anthropic didn't take this seriously enough. If Anthropic had treated it seriously, instead of downplaying it as an isolated event, and proactively fixed it or suspended access, things never would have escalated to this point."

Another person close to Anthropic countered that the so-called "jailbreak" did not mean Fable 5's safety system had failed, and noted that the company had communicated with the government before releasing Fable. This person stated that during multiple rounds of communication, the government did not object to Fable's release.

After the government imposed export controls, Anthropic stated in a blog post that it would comply with the government's directive, but believed the action was an overreaction.

Anthropic stated: "As we have said publicly, we believe the government should have the power to prevent unsafe deployment, but this should be based on a transparent, fair, and clear legal process grounded in technical facts. This action does not meet those principles."

A White House official, speaking on condition of anonymity, said innovation remains the administration's "top priority, but we must also prioritize safety."

Amazon declined to disclose details of its communications with the government in its statement. An Amazon spokesperson said: "It is not uncommon for the government to seek our input on potential security risks. When such situations occur, we do not disclose the details of those discussions."

In early April, Anthropic announced that its latest powerful model, Mythos, would only be made available to a small number of technology and cybersecurity companies to help them test software vulnerabilities. Anthropic stated at the time that due to the model's significant capabilities, it could cause severe consequences if it fell into the wrong hands, necessitating a restricted release.

The unveiling of Mythos led to a series of meetings between Amodei and senior White House officials. Both sides described these meetings as productive. Subsequently, multiple rounds of discussions ensued regarding how to regulate advanced AI models, eventually contributing to a recent executive order requiring companies to voluntarily submit advanced models for government review before widespread deployment.

Fable 5, released publicly this week, was described by Anthropic as a "Mythos-class model" but equipped with additional safety measures to allow it to be opened to general users. The model had been reviewed by the U.S. government and the UK AI Safety Institute.

However, after the alleged safety flaw was disclosed, multiple government officials concluded the model needed to be taken down immediately.

On Saturday morning, David Sacks, former White House AI director and a staunch opponent of regulation, posted on X supporting the government's export controls on Anthropic.

Sacks stated he did not believe the "jailbreak" was simple or trivial, nor did he think the export controls were an attempt by the government to broadly control the AI industry. Sacks wrote: "The government now wants Anthropic to fix the safety issue, then lift the export controls, and allow Fable to return to public release. The government wants this to happen as quickly as possible. Frankly, the government is puzzled that Anthropic, which has always said safety requests are its highest priority, is now unwilling to cooperate."

Sacks and other officials within the administration have previously criticized Anthropic, accusing it of left-wing political bias and alleging that the company was spreading fear by advocating for stronger industry regulation and warning of massive job displacement.

Earlier this year, the rift between the government and Anthropic reached a new high with Defense Secretary Pete Hegseth and the Pentagon. On March 3, because Anthropic refused to allow its AI tools to be used for mass domestic surveillance and autonomous weapons systems, the Pentagon designated the company as a supply chain risk.

On Saturday, Sacks stated that the pre-existing disputes between the government and Anthropic were separate from the export control decision. Sacks wrote: "The government values Anthropic's technical capabilities and believes this problem, while serious, should be easily solvable. The ball is now in Anthropic's court."

Original Link