The US government has granted Anthropic the green light to release its advanced Mythos AI model to a select group of companies and organizations, reversing previous export restrictions that were imposed earlier this month due to national security concerns.
Following the export ban in June, Anthropic engaged with the US government to mitigate risks related to its Covered Models, as detailed by Commerce Secretary Howard Lutnick in a letter to the company issued on Friday.
Lutnick noted that, given the progress made, "I have determined that appropriate safeguards are in place to permit certain trusted partners to access the Claude Mythos 5 Model."
Notably, the letter did not provide authorization for Anthropic to distribute Fable, a less powerful variant of Mythos.
In a statement, Anthropic confirmed that they received notification from the US government allowing the deployment of Mythos 5, touted as their most robust cybersecurity model, to a limited group of cybersecurity defenders and infrastructure providers.
The company expressed its commitment to quickly set up the approved providers and restore their access to Mythos 5, stating, “We are pleased to see this progress and will continue to collaborate with the government to broaden access to Mythos 5 and reintroduce Fable 5 for general use.”
Discussions between Anthropic and the government are anticipated to continue over the weekend, especially concerning the potential restoration of access to Fable, according to a source familiar with the matter who spoke to CNN.
Earlier this month, Anthropic had to suspend customer access to both Mythos and Fable in compliance with the government’s order to halt all use by foreign nationals, including its own employees.
This export ban and the subsequent negotiations highlight the current lack of a cohesive regulatory framework surrounding AI technology, particularly as advancements are rapidly unfolding and the US aims to maintain its competitive edge against nations like China.
Semafor was the first to report on the government's approval for the release of Mythos.
Until recently, the Trump administration had favored a largely hands-off regulatory stance on artificial intelligence, primarily to keep US firms ahead of their Chinese counterparts. However, the recent actions regarding Anthropic, along with requests to OpenAI to limit its latest model, stem from longstanding concerns among experts that Mythos could pose a security threat by enabling cybercriminals to swiftly identify and exploit vulnerabilities using sophisticated AI.
Furthermore, the relationship between the federal government and Anthropic has encountered additional challenges. Earlier this year, the administration categorized Anthropic as a "supply chain risk," effectively placing it on a blacklist due to disagreements surrounding the military applications of its products. Anthropic has since initiated legal action against this classification and has already achieved at least one initial success in the ongoing litigation.
