AI
US Government Orders Anthropic to Suspend Fable 5 and Mythos 5 AI Models Over National Security Concerns
The US government has ordered Anthropic to suspend access to its Fable 5 AI model over national security concerns, impacting users worldwide.
Estimated Reading Time: 3 minutes
- The US government has ordered Anthropic to suspend access to its AI models, Fable 5 and Mythos 5, for all foreign nationals due to national security concerns.
- Anthropic issued a statement challenging the government’s decision, asserting that concerns stem from a minor vulnerability linked to a “jailbreak” technique.
- This suspension affects all Anthropic customers, including foreign-national employees.
- The directive highlights escalating US national security measures regarding advanced technology exports.
- Anthropic is seeking to restore access and comply with the regulations while addressing the underlying issues.
Context / Background
Anthropic, a leading company in artificial intelligence development, recently released the advanced models, Fable 5 and Mythos 5. According to a report from the Times of India, the directive to suspend access was received by Anthropic on a Friday evening at 5:21 p.m. ET, leading to an immediate response from the company to disable both models for all customers, a move that encompasses not only foreign users but all customers, including Anthropic’s foreign-national employees.
Key Details
The US government’s export control order specifically targets Fable 5 and Mythos 5, articulating concerns that are tied to national security logistics. As part of compliance with this directive, Anthropic is disabling the models for all customers because maintaining access could risk non-compliance with the order. Notably, the order does not affect access to Anthropic’s other AI models.
Anthropic contends that the national security concerns referenced by the US government relate to a “jailbreak” technique purportedly shared with officials, which allegedly exposes vulnerabilities in the models. However, the company argues that these vulnerabilities are minor and have been documented in other AI systems as well.
Impact
The suspension of Fable 5 and Mythos 5 affects Anthropic’s operations significantly, disrupting service for all its customers. Anthropic has expressed regret over this situation and aims to restore access “as soon as possible,” while clarifying that it views this suspension as a temporary measure.
The directive underlines a tightening of US national security measures regarding the export of advanced technologies, especially AI systems, to foreign nationals. This could have significant implications for global AI collaboration and research, particularly affecting users in various jurisdictions, including India.
What’s Next
Anthropic has indicated its intention to comply with the US government’s directive while reiterating its concerns regarding the rationale behind it. As this issue unfolds, the company’s ability to effectively communicate technical details to regulators will be critical. The broader implications of such regulations on AI models could lead to increased scrutiny and potential re-evaluation of export control policies related to emerging technologies.
In light of Anthropic’s response and the subsequent discussions about the nature of AI vulnerabilities, further communication between the company and the US government may pave the way for clarifications that could help in restoring access to both models in the near future.
FAQ Section
Why did the US government order the suspension of Fable 5 and Mythos 5?
The US government cited national security concerns associated with vulnerabilities in the AI models, specifically those potentially exposed through a “jailbreak” technique.
How does this affect Anthropic’s customers?
The suspension affects all Anthropic customers globally, including foreign-national employees, resulting in a temporary loss of access to these AI models.
What are Anthropic’s next steps?
Anthropic aims to comply with the directive while working to address the concerns raised by the government to potentially restore access to the models.