Last weekend Anthropic sent some technical managers to Washington to discuss with government officials, after the Donald Trump administration imposed the shutdown of two advanced artificial intelligence models – Fable 5 and Mythos 5 – for foreign entities and individuals. Given the scope of the order, however, the company decided to block the two models for all users.
WHAT IS THE “PROBLEM” WITH ANTHROPIC’S FABLE 5 AND MYTHOS 5?
The government intervention on Fable 5 and Mythos 5 is related to the capabilities of the two models – which belong to the same group, called Mythos – to identify system vulnerabilities with particular accuracy, potentially allowing them to be exploited for attacks. Anthropic has limited the use of Mythos to a handful of companies – including Amazon Web Services, Microsoft, Apple, Google, and Nvidia – that have agreed to use it exclusively for defensive purposes. In recent days, however, it publicly released a “downgraded” version of the model: Fable 5, precisely.
On June 2, with an executive order, President Trump had asked AI companies to deliver their most advanced language models to the government so that it could test them before release.
WHAT DOES AMAZON HAVE TO DO WITH IT
The government intervention on Fable 5 and Mythos 5, which then led to the total blocking of the two models, originated from a report by Amazon, which had identified a security flaw in Fable 5. The CEO of the tech company, Andy Jassy, communicated the discovery to several White House officials, including Treasury Secretary Scott Bessent.
According to Anthropic, however, the vulnerability of Fable 5 reported by Amazon – which has free access to Mythos – is not that significant. In a statement, it said it “does not believe that identifying a potential security vulnerability of limited scope should constitute a valid reason to withdraw from the market a commercial model already distributed to hundreds of millions of people.”
The startup also believes – as reported by Quartz – that the demonstrated capability level is already present in other publicly available models, including OpenAI’s GPT-5.5, and that it is routinely used by cybersecurity experts.
In April, Amazon had announced an investment of up to 25 billion dollars in Anthropic.
HOW IS ANTHROPIC DOING: CONFLICTS WITH THE PENTAGON AND IPO PLANS
The blocking of Fable 5 and Mythos 5 came at a complicated time for Anthropic. Relations with the American government were already difficult, after the Department of Defense had classified the company as a “supply chain risk” following some disputes over the use of its models for military purposes.
Anthropic is also preparing for the initial public offering (IPO, in jargon) and recently reached a valuation of 965 billion dollars, higher than that of OpenAI, its main rival. In early June Anthropic filed an IPO prospectus with the SEC (the US federal agency that oversees the stock market), declaring a revenue run rate (a projection of annual revenues based on current performance, simplifying) of 47 billion dollars.
