Artificial Intelligence—A deadly new arms race

The US and China are vying for technological and economic supremacy (Photo: WikipediaCommons)

Somewhere­ deep in a ­laboratory in San Francisco, California, sits an Artificial Intelligence model so ­powerful that its locked in a digital vault.

They say it can break into any computer system, tear apart a country’s defences, shut down its power ­stations and loot its banks­. Only those working for the US state will be allowed to access it, they insist.

Anthropic’s Claude Mythos was launched last month and is AI’s most able Large Language Model (LLM) yet. LLMs are ­generally ­data‑mining tools that repackage existing information.

But according to the Financial Times ­newspaper, Claude Mythos is “the first AI model that can ­autonomously discover, chain together and exploit or patch software vulnerabilities more ­effectively than almost every human researcher, at an unprecedented scale.”

AI in this context is not a productivity tool, it is a weapon. No wonder that Russian state media describes Mythos as equivalent to the “atomic bomb”.

We are now in an AI arms race.

China, as the only world power with the potential to compete with the US, is racing to catch up. It knows every aspect of its state machine is vulnerable—and that it has no ability to retaliate.

Donald Trump wants to keep it that way. For years, US ­officials have estimated that AI models developed by its most advanced firms are some six months ahead of China’s best models.

Now, they plan to build AI supercomputers across the US in a project called Stargate, which involves the firms OpenAI, Oracle and Microsoft.

While the project is ­nominally privately funded and controlled, Stargate itself was launched in the White House. Trump pledged to use executive orders to fast track the energy and infrastructure it needs.

The British establishment’s Chatham House think tank, notes, “Stargate’s scale hints at the growing convergence of national security, infrastructure and Silicon Valley interests.”

Trump’s supporters want to further press their ­advantage by acting against the free market that capitalists ­normally ­trumpet. They demand the US halt all exports of AI chips and semiconductor ­manufacturing equipment to China.

In response, the Chinese state has boosted ­massively its own semiconductor design and ­manufacturing industry, which it has sealed-off from the global market.

The state is directly ­subsidising firms such as Huawei and Cambricon to develop ­alternatives to US chips.

Whether Chinese-made AI models will be able to keep up remains to be seen. But the state has already hit back ­effectively by blocking exports of ­gallium and germanium, ­critical ­materials in chip making.

China’s fightback is ­worrying the Trump administration. “If we don’t win in AI, then it’s game over,” the US treasury secretary told the Wall Street Journal newspaper last month.

It’s easy to see why atomic rhetoric and Cold War ­analogies have purchase. But they can make it feel as though the battle over AI is purely the result of commpetition between rival stat­es.

But the way giant AI ­corporations are increasingly intertwined with the state reveals something far more ­systemic than national or ­corporate competition.

In the Marxist tradition, ­imperialism is not just one state dominating other states. It comes directly from the ­development of ­capitalism—a system based on the exploitation of labour and driven by the ­competitive ­accumulation of capital.

One pioneer of this theory was the German socialist Rudolf Hilferding.

In his 1910 book, Finance Capital, he explained that ­capitalism would lead to a concentration and ­centralisation of capital into the hands of fewer and fewer capitalists. This would then lead to a growth in monopolies and competition between these capitalists.

Here major firms’ needs and the needs of the state become interwoven. All talk of free ­markets and the rule of law become subsumed in something far bigger—the “national interest.”

Ultimately, every developed capitalist state would become locked into global competition in order for its domestic ­economy to ­compete with the ­monopolies associated with other states.

As Hilferding put it, “Finance capital does not want freedom, but domination. It has no regard for the ­independence of the ­individual capitalist, but demands the state should be strong enough to pursue its interests everywhere.”

Today, these major firms are the AI and tech giants that have subsumed rivals to become “too big to fail.”

In 1916, the Russian Marxist Vladimir Lenin took the theory of imperialism further, ­arguing that the reorganisation of ­capital into monopolies was the system’s final, parasitic stage.

He argued that giant ­conglomerates that combined with the state did more than trade, they divided the world into spheres of influence and prepared the ground for war.

“This is a new stage of world concentration of capital and production, incomparably higher than the preceding stages,” he wrote.

“Let us see how this ­super-monopoly develops. They first divided the home market among themselves. But the home market, under capitalism, is inevitably bound up with the foreign market.”

To dominate the “foreign market,” said Lenin, the ruling class would use the nation’s ­military as a threat against rivals.

Today’s tech and AI ­oligarchs once preached an ideology of a shrunken state, where taxes, pay and regulations would be so low that pioneering firms could flourish. Some even thought that corporations themselves would become the basis for a new “network state.”

Meta, the owners of Facebook and Instagram, for ten years planned to build a 59-acre town called Willow Village near its headquarters in Menlo Park, California.

Before it was cancelled ­earlier this month, Willow Village was to be Meta’s own mini state, free from some of the “constraints” limiting other towns.

But intensifying ­competition is developing between the US and China. Fearful of China emerging as a competitor on the global stage, the US is competing to hold onto its economic and technological supremacy.

Most tech bosses have learned to love the ­militarised ­international power that only the ­capitalist nation state provides.

As well as protection at home and abroad, bosses also need the state for financial support because they are in ­competition with other firms fed on government subsidies.

That’s been a big part of tech firms’ recent pivot away from talk of being for “public benefit.”

Google recently removed stipulations preventing its ­products from being used to create weaponry. It now allows the military to embed Gemini AI technology in its programs.

OpenAI had a ban ­similar to Google’s but since 2024 it has had a serving US lieutenant colonel on its board. Now, together with other tech firms, it has signed up to a $200 million contract with the Department of War.

For a short time, Anthropic appeared to be an outlier. It even sued the Department of War for blacklisting the firm when it refused to allow its AI to be used for domestic mass ­surveillance and fully ­autonomous weapons.

But that’s changing now.

Co-founder and chief ­executive Dario Amodei ­emphasises, “Anthropic has much more in common with the Department of War than we have differences.”

And in a blog post in January, he warned of the deadly threat of bioweapons developed by Chinese AI. But he ­simultaneously argued that firms such as his should arm democratic governments with the most advanced AI possible to guard against that danger.

It came as no surprise to learn that the US military used Anthropic’s Claude AI to select targets in Iran. Perhaps the Shajareh Tayyebeh school in Minab was one of them. If so, then the blood of more than 100 children is on Amodei.

Many tech workers are enraged at the way their labour is being used for war, rather than improving people’s lives.

In 2018, over 3,000 Google workers signed a petition against the firm’s involvement in a government project to use their AI to analyse military drone footage.

A few years later, Google workers protested against the firm for allowing its products to be used in Israel’s genocide of the Palestinians.

This year, Google’s DeepMind AI workers in Britain voted to unionise partly in response to the company ­signing a new ­contract with the Pentagon.

Microsoft and Amazon ­workers have also acted against bosses that talk of “doing good” all the while profiting from war.

The hypocrisy of the ruling class—in both its industrial and state forms—is a terribly ­familiar story. Writing months before the carnage of the Second World War, Russian revolutionary Leon Trotsky issued a warning.

“War is a gigantic ­commercial enterprise, especially for the war industry,” he wrote.

“The ‘60 Families’—the ruling financial and ­industrial elite are therefore ­first-line ­patriots and the chief ­provocateurs of war. Workers’ control of war industries is the first step in the ­struggle against the ‘­manufacturers’ of war.”