Inside the Whirlwind 24 Hours That Led the White House to Slap Export Controls on Anthropic
However, administration officials remained unconvinced, citing concerns that the AI model's security measures were inadequate. The officials pressed Amodei to voluntarily withdraw Fable from public access, warning that failure to do so could lead to more drastic measures. Amodei reportedly pushed back, arguing that Anthropic had taken sufficient precautions to prevent the model from being misused. Despite the back-and-forth, the administration ultimately decided to impose export controls on Anthropic, effectively forcing the company to pull Fable from the market. The move marked a significant escalation in the government's efforts to regulate the rapidly evolving AI landscape. As the situation continued to unfold, questions swirled about the implications of the export controls and the potential consequences for the broader tech industry. The administration's actions also sparked debate about the appropriate balance between innovation and national security in the development of advanced AI models. Meanwhile, Anthropic's decision to comply with the export controls raised concerns among some experts about the potential chilling effect on AI research and development.
Amodei's defense of Fable's security measures was met with skepticism by Cairncross and Bessent, who remained unconvinced by his arguments. The administration's concerns were further validated by Amazon's findings, which were reviewed by the National Security Agency. According to a White House official, the NSA's assessment provided "proof" of the potential risks associated with Fable.
Following the imposition of export controls, Anthropic expressed its disappointment, stating that the move was disproportionate and did not align with the principles of transparency, fairness, and technical facts. The company emphasized its commitment to complying with the government's directive while reiterating its concerns about the potential impact on AI research and development.
The White House, however, maintained that national security considerations took precedence over innovation. A senior official, speaking on the condition of anonymity, underscored the administration's commitment to balancing security and innovation. "Innovation is our number one goal, but we also have to prioritize security," the official said.
As the debate surrounding Fable's security continued, it emerged that Amazon was not the only company to have raised concerns about the model. Three people familiar with the government's thinking revealed that other companies had also flagged potential issues, citing Anthropic's perceived lack of seriousness in addressing the vulnerabilities. According to one of these individuals, "Had Anthropic taken it seriously and moved to fix or pause access, this would have never happened."
Amazon declined to elaborate on its discussions with the administration, with a spokesperson stating that it's not uncommon for governments to seek counsel on potential security risks, and that the company doesn't share details of these discussions.
Anthropic's latest model, Mythos, was initially only available to a limited set of tech and cyber firms due to its powerful capabilities, which the company said could be misused if it fell into the wrong hands. The model's release led to a series of meetings between Anthropic's representatives and senior White House officials, resulting in conversations about regulating advanced models.
These conversations culminated in a recent executive order, which encouraged companies to voluntarily submit their advanced models to the government for review before deploying them widely. Anthropic's Fable 5 model, described as a "Mythos-class model" with built-in safeguards, was reviewed by the administration and the UK's AI Security Institute before its public launch.
However, after alleged security flaws were disclosed, administration officials felt that the model needed to be pulled from public access. David Sacks, the former White House AI czar, expressed agreement with the administration's decision to impose export controls on Anthropic, stating that he did not believe the reported "jailbreak" was simple or not serious.
Sacks noted that the administration's goal is for Anthropic to address the safety issue, have the export control lifted, and allow Fable to return to general release as soon as possible. He criticized Anthropic for not complying with safety requests that it had previously stated were its top priority.
The administration has been critical of Anthropic, accusing it of having a leftist political bias and fearmongering due to its advocacy for stronger industry regulation and warnings about job disruption. Earlier this year, the Pentagon designated Anthropic a supply chain risk due to its refusal to allow its AI tools to be used for mass domestic surveillance and autonomous weapons.
Sacks downplayed the significance of past feuds between the administration and Anthropic, stating that the export control decision was a separate issue and that the administration values Anthropic's technical capabilities. He emphasized that resolving the issue is now in Anthropic's hands.
Amodei's defense of Fable's security measures was met with skepticism by Cairncross and Bessent, who remained unconvinced by his arguments. The administration's concerns were further validated by Amazon's findings, which were reviewed by the National Security Agency. According to a White House official, the NSA's assessment provided "proof" of the potential risks associated with Fable.
Following the imposition of export controls, Anthropic expressed its disappointment, stating that the move was disproportionate and did not align with the principles of transparency, fairness, and technical facts. The company emphasized its commitment to complying with the government's directive while reiterating its concerns about the potential impact on AI research and development.
The White House, however, maintained that national security considerations took precedence over innovation. A senior official, speaking on the condition of anonymity, underscored the administration's commitment to balancing security and innovation. "Innovation is our number one goal, but we also have to prioritize security," the official said.
As the debate surrounding Fable's security continued, it emerged that Amazon was not the only company to have raised concerns about the model. Three people familiar with the government's thinking revealed that other companies had also flagged potential issues, citing Anthropic's perceived lack of seriousness in addressing the vulnerabilities. According to one of these individuals, "Had Anthropic taken it seriously and moved to fix or pause access, this would have never happened."
Amazon declined to elaborate on its discussions with the administration, with a spokesperson stating that it's not uncommon for governments to seek counsel on potential security risks, and that the company doesn't share details of these discussions.
Anthropic's latest model, Mythos, was initially only available to a limited set of tech and cyber firms due to its powerful capabilities, which the company said could be misused if it fell into the wrong hands. The model's release led to a series of meetings between Anthropic's representatives and senior White House officials, resulting in conversations about regulating advanced models.
These conversations culminated in a recent executive order, which encouraged companies to voluntarily submit their advanced models to the government for review before deploying them widely. Anthropic's Fable 5 model, described as a "Mythos-class model" with built-in safeguards, was reviewed by the administration and the UK's AI Security Institute before its public launch.
However, after alleged security flaws were disclosed, administration officials felt that the model needed to be pulled from public access. David Sacks, the former White House AI czar, expressed agreement with the administration's decision to impose export controls on Anthropic, stating that he did not believe the reported "jailbreak" was simple or not serious.
Sacks noted that the administration's goal is for Anthropic to address the safety issue, have the export control lifted, and allow Fable to return to general release as soon as possible. He criticized Anthropic for not complying with safety requests that it had previously stated were its top priority.
The administration has been critical of Anthropic, accusing it of having a leftist political bias and fearmongering due to its advocacy for stronger industry regulation and warnings about job disruption. Earlier this year, the Pentagon designated Anthropic a supply chain risk due to its refusal to allow its AI tools to be used for mass domestic surveillance and autonomous weapons.
Sacks downplayed the significance of past feuds between the administration and Anthropic, stating that the export control decision was a separate issue and that the administration values Anthropic's technical capabilities. He emphasized that resolving the issue is now in Anthropic's hands.
#News, #USA