AI Models By China, Japan May Have Led To US Lifting Mythos Ban

The US government has lifted its export ban on Anthropic's most powerful AI models - Mythos 5 and Fable 5. Several experts believe there may be a very strong Asia angle to it.

During the ban Japan and China both released AI models which as per some benchmark tests reportedly matched Mythos in certain capabilities such as cyber security.

These were Tulonfeng from China's 360, a cyber security major and 
Zhipu AI's model GLM-5.2, also from China, along with Sakana AI's Fugo Ultra from Japan.

Sakana said its frontier AI model "stands shoulder-to-shoulder with leading models like Anthropic's Fable 5 and Mythos Preview."

Also read: A Japanese AI System Reportedly Beat Claude 5 On Certain Benchmarks

As far as Mythos' capabilities are concerned, Anthropic Co-founder and CEO Dario Amodei himself admitted that some early testers had told the company that it was so powerful, it should require a "gun license" to use. When Mythos was first previewed it was able to identify flaws in critical systems that had remained undetected for decades.

Also read: Mythos A Weapon, Needs Gun License, Testers Warned Anthropic

The US would, of course, hate to lag behind in the AI race and most experts believe that the Asians catching up may have played a very strong role in lifting the ban.

AI expert and CTO, AiEnsured Srinivas Padmanabhuni told NDTV that he believes its "very much" one of the reasons that pushed the US to lift the ban.

However, among AI analysts there is skepticism around whether the Asian models are truly at par with Mythos.

"Reportedly at par but not exactly 'at par' as we talk about Fugu it's a layer, routing other models API (Application Programming Interface) than trained from scratch and even Tulongfeng numbers were their own, rather than independently bechmarked," Sagar Vishnoi, co-founder of Future Shift Labs told NDTV.

He was quick to add, that despite his skepticism, he believes that the Asian models are definitely good enough, even if not truly at par.

"I still believe these models could have filled the vacuum in near future filling up the space for Mythos and Fable," Vishnoi said.

Anthropic had been in dialogue with the White House ever since the Trump administration last month said that no foreigner whether in the US or outside could have access to these models on account of national security, even if they were Anthropic employees. To comply, the company was forced to revoke access for everyone.

A Complicated Relationship

There's also a strong school of thought that the Trump administration was arm-twisting Anthropic to show who's really the boss.

In April, the Pentagon labelled Anthropic a "supply chain risk" after the company refused to accept contract language the US government was insisting upon. Then around mid-June, it slapped the export ban on the company after the alleged discovery of a Fable 5 jailbreak that could potentially allow users to circumvent guardrails designed to prevent bad actors such as hackers from accessing Mythos' full capabilities. 

Also read: Why Anthropic Disabled Its Most Powerful AI Model Days After Launch

Anthropic had maintained that its guardrails were solid, and what the US government was trying to point to was already present in existing models of rival AI companies.

Anthropic and the US government have had a rather complicated relationship. It has been locked in a standoff with the Trump administration for refusing to allow its technology to potentially be used for mass surveillance and autonomous weapons, leading the Pentagon to cut contracts with the company.

In fact, during the recent round of talks with the White House to get Mythos back up, officials reportedly were of the opinion that Amodei is a "weirdo" who's difficult to deal with. That's when Tom Brown, the other co-founder of Anthropic stepped in and seemingly took the talks to the finish line.

Ahead of Trump's re-election campaign Amodei called Trump "a feudal warlord" in a Facebook post (he later deleted), while urging people to vote for rival Kamala Harris. 

When the US government labelled Anthropic a "supply chain risk" Trump had posted on social media platform Truth Social that he believed the company consisted of "leftwing nut jobs" who were trying "strong-arm" the Pentagon.

US defense secretary Pete Hegseth, had called Amodei "an ideological lunatic" during a congressional hearing.

We may not have seen the last of the Anthropic-White house relationship drama, but for now they seemed to have buried the hatchet at least temporarily.