Fable5被迫下线前24小时，白宫发生了什么？

Original title: 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 editor: Peggy

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

Behind this sudden takedown was a high-stakes 24-hour power struggle between the White House and Anthropic. Concerning whether Fable 5's safety guardrails could be bypassed, Anthropic CEO Dario Amodei held several tense phone calls with senior government officials, including the Treasury Secretary, the Commerce Secretary, and the White House Cyber Director. The government believed the model could be used to identify software vulnerabilities, posing a national security risk. Anthropic countered that the issues were not a "general jailbreak" and criticized the government's lack of a transparent, clear, and fact-based legal process for its actions.

The crux of this incident extends beyond a single model takedown or a breakdown in communication between the White House and an AI company. It signifies that advanced AI models are being redefined as potential national security assets. As model capabilities penetrate sensitive domains like cybersecurity, vulnerability discovery, and intelligence analysis, the government is no longer satisfied with company self-assessments and voluntary reviews. It is resorting to more aggressive policy tools, such as export controls, to directly intervene in the pace of model releases.

Ironically, Anthropic has been one of the most vocal advocates for AI regulation within the industry, yet it has now become the target of the regulatory hammer. The government believes its response to the security flaw was insufficiently serious, while Anthropic views the White House's reaction as an overreach. Underlying this dispute is a new question the AI industry must face: Who has the authority to determine if a model is safe enough? Is corporate self-certification of safety still sufficient? And how will rapid interventions justified by national security alter compliance costs, release schedules, and the global competitive landscape for AI companies?

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

Below is the original article:

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

According to two government officials and a senior White House official, during this turmoil, Anthropic CEO Dario Amodei held several tense phone calls with top government figures, including Treasury Secretary Scott Bessent and White House Cyber Director Sean Cairncross. These individuals spoke on condition of anonymity as they were not authorized to discuss the matter publicly.

This incident also brings a pressing question to the forefront: With advanced AI models evolving rapidly and their capabilities constantly expanding, how should the White House draw the line between innovation and safety?

The details of these calls have not been previously reported.

The Night Before the Takedown: A Dispute Over Safety

Following the government's imposition of 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 the model was safe and controllable, but shortly after its release, senior government officials began to reassess whether its safety guardrails were as robust as the company claimed.

Two government officials and a senior White House official stated that on Thursday, two days after Fable's public launch, Amazon CEO Andy Jassy expressed concerns to the White House that the model's safety guardrails could be at risk of being bypassed.

A person familiar with Amazon's communications said the company 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 government officials and a senior White House official, Bessent, Cairncross, White House Chief of Staff Susie Wiles, and other senior officials met to discuss the model and the government's response. One of the sources said Bessent participated remotely as he was traveling 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 attending a wellness retreat, according to a government official and a senior White House official.

An Anthropic spokesperson denied this claim, stating: "That is completely untrue."

A person close to Anthropic said the government first requested to speak with Amodei around noon, and Amodei was on the phone with senior officials within 1 hour and 15 minutes. The person stated that during the time 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. A senior White House official and one government official said roughly half a dozen senior government officials were on the calls, including Cairncross, Bessent, and Commerce Secretary Howard Lutnick.

The senior White House official added that other White House staff and government officials participated in some calls, 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.

During the calls, Amodei attempted to clarify what he believed were misunderstandings. He pushed back against the government's safety concerns, defended Anthropic's protective mechanisms, and emphasized that the bypass methods already discovered were specific, contextual issues, not a broader "jailbreak" – meaning the model had not been completely freed from Anthropic's safety guardrails.

In a blog post published after the export controls were imposed, Anthropic stated: "No testers so far have found a general jailbreak method – one that would circumvent the model's safety measures on a broad scale, unlocking a wide range of cyber capabilities." The company also said it is currently unrealistic for Anthropic, or any other AI company, to completely prevent all jailbreak attempts.

Anthropic also defended its safety systems, stating its guardrails are "so strong that many users complain they are too broad."

However, Cairncross and Bessent were not convinced by Amodei's explanations. A White House official said Amazon's findings were submitted to the National Security Agency for review, which considered the material 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 vulnerabilities. Amodei requested more time and information but did not commit to taking the model down. The senior White House official said that during one call, Bessent directly told Amodei he was making a "bad decision."

The Regulatory Hammer Falls: Fable 5 and Mythos 5 Face Controls

Shortly after the calls ended, the Trump administration imposed export controls on both the Fable 5 and Mythos 5 models. Anthropic stated the government invoked national security authority to prohibit foreign nationals from using these models. The company said the "practical effect" of the order was that, to ensure compliance, Anthropic had to "abruptly disable" access to the relevant models for all customers.

A senior White House official said: "We pleaded with them to work with us for hours. The export controls were a last resort. This is not something we wanted to do, but we saw no other option."

Following the report, a person close to Anthropic pushed back, saying the company was not given a choice for "voluntary cooperation."

The person said: "The White House gave us only 90 minutes to take the model down, without providing any details about the actual threat. There was never any pleading or request to work with us, just an announced 90-minute deadline."

A senior White House official said White House officials had heard Amodei compare the dangers of Anthropic's technology to a nuclear bomb. Therefore, when the CEO, who had consistently advocated for AI safety regulation, was reluctant to take down a system to fix known security vulnerabilities, government officials found it difficult to understand.

Anthropic has long been seen as one of the industry's most vocal advocates for AI regulation. The company has repeatedly argued for regulatory frameworks to address the potential global safety risks and job displacement stemming from the rapid development of AI.

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

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

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

After the government implemented the export controls, Anthropic stated in a blog post that it would comply with the government directive, but argued the move was an overreaction.

Anthropic said: "As we have said publicly, we believe the government should have the authority to prevent unsafe deployments, but this must be based on a transparent, fair, 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 in a statement to disclose the specifics of its communications with the government. An Amazon spokesperson said: "It's not unusual for the government to seek our input on potential security risks. When this happens, we do not disclose the details of those discussions."

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

Mythos' launch led to a series of meetings between Amodei and senior White House officials. Both sides described these meetings as productive. This led to further discussions on regulating advanced AI models, eventually culminating in a recent executive order requiring companies to voluntarily submit advanced models for government review before widespread deployment.

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

However, after the so-called safety flaws were disclosed, multiple government officials believed the model needed to be taken offline immediately.

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

Sacks stated he did not believe the "jailbreak" was simple or inconsequential, nor did he believe the export controls were an attempt by the government to broadly control the AI industry. Sacks wrote: "The administration now wants Anthropic to fix the security problem, then lift the export controls and return Fable to public release. The administration wants this to happen as quickly as possible. Frankly, the administration is puzzled that Anthropic, which has always said safety requests were its top 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 of creating panic by advocating for stronger industry regulation and warning of massive job displacement.

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

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

Original Link