Washington Holds the Pen, Mythos Provides the Ink

Artificial intelligence has ushered in a new age of geopolitics. The unveiling of Anthropic's Mythos AI model in April may one day be remembered as a historic turning point. But this is not a technology story. It is a geopolitical one. How the technology is being distributed, and how Washington has responded to its unveiling, tells us a great deal about the world that is emerging. New geopolitical lines are being drawn, and Mythos may prove to be the ink.

What is it exactly? Good question.

Mythos is an AI system described as being “strikingly capable” at computer security tasks. In simple terms, it can identify vulnerabilities far more effectively than humans. Unsurprisingly, it has not been released to the public. Its capabilities are simply too powerful and the risks too obvious. Following its launch, Anthropic CEO Dario Amodei met with White House officials to discuss federal deployment and the pace of any international rollout.

However, the technical workings of the model are not the focus here. Its political significance is. For years, frontier AI was discussed largely through the lens of productivity and consumer applications. Mythos may represent the clearest example yet of advanced AI being treated as a national security asset. Like intelligence capabilities or advanced military systems, its value lies not only in what it can do but in who controls it.

Washington has been handed a new superpower.

Traditionally, a breakthrough of this magnitude would be shared among close allies, particularly within NATO and the Five Eyes intelligence partnership. But times have changed. And how they have changed. The White House is now occupied by an administration that has shown little interest in maintaining many of the assumptions that underpinned post-war alliances. 

Since Mythos was unveiled, only American banks have been allowed to test their systems against the model. The result is a natural asymmetry between the American financial system and its international counterparts. Even now, months after the launch, that gap remains. The irony is hard to miss. European regulators have been warned about the rise of AI powered cyber threats while being denied access to one of the most advanced tools available to understand and defend against them.

Europe has not remained silent. Officials across the continent have increasingly voiced concerns about the rollout. In Britain, those concerns have become particularly pointed. Speaking at a Central Banking Conference in Reykjavik, Bank of England Governor Andrew Bailey confirmed that UK banks still had not been granted access. The reason, he claimed, was that it had been “caught up in the process with the US administration”. 

Those comments are striking.

Britain is not just another ally. It is a founding member of Five Eyes, Washington's closest intelligence partner and a global financial centre. If the UK is struggling to gain access, what does that say about the future of technology sharing between allies?

As of early June, Britain still had not been granted access despite other European countries, including France, Germany and Italy, beginning to receive the technology. The most surprising omission from Anthropic's international rollout may not be China or Russia, but Britain itself.

That should concern policymakers in London.

For decades, Britain has operated on the assumption that close security and intelligence ties with Washington would guarantee access to strategically important technologies. Mythos suggests that assumption can no longer be taken for granted. Access to frontier AI has become a matter of strategic power, and even America's closest allies can no longer assume they will automatically be included.

Yet the treatment of allies is only half the story.

China and Russia have, as expected, been excluded entirely. If Mythos were being treated as a normal commercial product, Anthropic would have every incentive to sell access to one of the world's largest technology markets. Instead, access has been restricted to a carefully managed network of trusted governments, institutions and infrastructure operators.

The logic mirrors Washington's semiconductor strategy. Since 2022, the US has restricted China's access to the world's most advanced AI chips, including Nvidia's A100 and H100 processors, while adding 140 Chinese entities to export control lists in 2024 alone. The goal is clear. Access to transformative technologies is no longer being governed solely by market capital, but by geopolitics. 

Throughout modern history, access to transformative technologies has shaped the international order. Nuclear weapons defined the strategic lines and hierarchy of the Cold War. Advanced semiconductors became central to competition between the US and China. AI is following suit. 

Mythos may ultimately be remembered less for what it could do than for how it was shared. Its rollout has offered a glimpse into the geopolitics of the AI age, where access to the most powerful systems is determined not by markets but by strategic interests. The delays faced by close allies, the selective expansion into Europe and the exclusion of rivals all point to the same conclusion. The US has not only unlocked a new superpower. It has begun deciding who gets to wield it.

Image: Flickr/Cory Doctorow

Licence.

No image changes made.