US AI arms race: Can America stay ahead of China as Pentagon struggles to control military AI risks?

AI is becoming a national security weapon as the US faces China in a new AI arms race, raising fears over cyberwarfare, weapons and AI safety.

America is entering a new AI arms race, but its own government is struggling to keep up. The US military, intelligence agencies and defense contractors are rapidly adopting artificial intelligence for cybersecurity, weapons, intelligence and national defense.

AI is becoming a national security weapon as the US faces China in a new AI arms race, raising fears over cyberwarfare, weapons and AI safety. (Pexel/Representative image) (Pexel)

The confusion became clear when the US Air Force first ordered military contractors to remove Anthropic's AI products. In mid-July, several of America's biggest defense contractors received a letter telling them that software used in weapons and control systems had to be free of Anthropic products by Sept. 1. The New York Times reported that companies could risk their Pentagon business if they failed to follow the order.

Pentagon changes its Anthropic AI ban

Then the Pentagon suddenly changed course. About a month later, the same contractors were told they did not have to immediately remove Anthropic products from Air Force systems and networks. The new guidance could still be reversed depending on the outcome of litigation.

Anthropic had wanted limits on two particularly sensitive uses of its technology. Amodei told Hegseth that Anthropic would not provide products to the military without assurances that they would not be used for domestic surveillance of Americans or for fully autonomous weapons operating without human control. The New York Times reported that the company's original 25 pages of restrictions were eventually reduced to these two conditions.

Mythos raises new cybersecurity risks

The ban created a new problem when Anthropic introduced Mythos, a powerful AI system designed to find weaknesses in computer software. Mythos can be used to identify and potentially exploit vulnerabilities in highly protected computer networks, making it especially important for cyberwarfare.

The National Security Agency, however, has continued testing Anthropic's advanced Mythos technology. NSA officials have argued that without Mythos, finding and fixing vulnerabilities in US military systems would become much harder. The New York Times reported that the NSA has been allowed to use versions of Mythos on an experimental basis despite the broader Pentagon restrictions.

The NSA is testing Mythos for both defense and offensive cyber operations. Officials said the agency is using the technology to test American military networks while also searching for weaknesses in networks belonging to countries including China, Russia, North Korea and Iran. The New York Times reported that critical tests are underway to understand whether Mythos could hack highly protected foreign networks.

That creates a major strategic problem for Washington: banning a powerful AI tool could leave America weaker against its rivals. One recently departed senior US official described stopping Mythos completely as similar to “unilateral disarmament,” because the technology could significantly change the cyberwarfare balance.

China is catching up in the AI race

AI is already being used across America's national security system for cybersecurity, intelligence and military operations, far beyond chatbots and everyday software. The bigger fear is that America's AI advantage may not last. US experts estimate that China could be only about six months or less behind America in developing the most advanced AI models. The United States still believes it has an advantage, but the gap could shrink quickly.

AI could make cyberattacks more powerful

China's progress matters because AI could make existing cyber threats much more powerful. Chinese hackers have already penetrated American communications and infrastructure networks in attacks known as Volt Typhoon and Salt Typhoon. That more advanced AI could amplify China's ability to attack communications networks, power grids and water systems.

Former Cybersecurity and Infrastructure Security Agency chief Jen Easterly said the basic threats existed even before AI became a major headline. What has changed, she argued, is that AI could dramatically increase the speed and scale of those attacks. Easterly made the comments to The New York Times.

US faces AI safety vs speed debate

The central argument is becoming a race between safety and speed. Supporters of faster AI development fear that stricter rules could cost America its roughly six-month lead over China, while critics worry that moving too quickly could allow dangerous systems to escape human control. The New York Times reported that this debate is now shaping US national security policy.

Washington has also struggled to create consistent rules for the most powerful AI models. Earlier this month, the Trump administration brought executives from major AI companies including OpenAI and Anthropic into discussions about reviewing new models for risks involving biological weapons, classified US systems and communications networks linked to nuclear weapons.

Open AI models create new security risks

But the government's policy has a major loophole: open AI models are exempt from the review process. These models can have their code customized by users, meaning they can potentially be modified for dangerous purposes. The New York Times reported that one technology executive briefed on the policy questioned why open models were excluded because they can also create serious risks.

This is especially important because many Chinese AI companies are developing open models. If America places heavy restrictions on its own private AI companies while Chinese companies continue developing customizable models, the policy could create a competitive disadvantage for the United States. That industry executives have raised concerns about this approach.

The Anthropic fight has also exposed a deeper problem: America may need the same AI companies it is trying to control. The Pentagon wants control over how military AI is used, while companies such as Anthropic want safeguards around applications they consider too dangerous. The New York Times reported that this conflict has become increasingly difficult to manage as AI capabilities advance.

Nvidia chips become part of AI race

America's AI policy has also been inconsistent when it comes to Nvidia's advanced chips. These chips are crucial because powerful AI models require enormous amounts of computing power to train and operate. The Trump administration has repeatedly changed its position on whether advanced Nvidia chips should be sold to China.

This has created a difficult balancing act for the White House. Restricting American AI companies too heavily could push foreign countries toward Chinese AI models, while weak regulation could increase the risk of dangerous AI systems being used against the United States. Chris McGuire, a senior fellow at the Council on Foreign Relations and former Biden administration official, described the administration's export-control policy as “incoherent,” according to The New York Times.

AI arms race looks like cold war

The situation is beginning to look similar to the early nuclear arms race, but AI is fundamentally different. During the Cold War, the United States initially had a nuclear weapons advantage and feared the Soviet Union would eventually catch up. CIA Director John Ratcliffe recently said the comparison between AI and “digital nuclear weapons” was not misplaced, according to The New York Times.

But AI cannot be controlled like nuclear weapons were. Nuclear technology was largely controlled by governments, while AI is being developed primarily by private companies and can spread quickly across borders. This makes AI arms control much more complicated than traditional weapons control.

AI is also changing far faster than traditional military technology. Some of the most powerful AI models are updated weekly, and researchers are often discovering new capabilities only after the systems are already being used. The New York Times reported that the speed and unpredictability of AI make traditional national security planning much harder.

China may be behind America in advanced semiconductor production even if it is catching up in AI models. US experts estimate that China could be months behind in developing leading AI models but several years behind America in producing the advanced chips needed to power them.

That semiconductor gap is one reason the US is using export controls as part of its AI strategy. Washington hopes limiting China's access to advanced chips will make it harder for Chinese companies to build the enormous computing infrastructure needed for frontier AI. The effectiveness of these restrictions remains heavily debated.

US struggles to control fast-moving AI

That silence highlights the larger challenge facing America: AI risks are arriving faster than governments can create rules for them. Military planners are dealing with AI systems whose capabilities can change within months, while government policies are changing within weeks. The New York Times' reporting on the Anthropic dispute illustrates the widening gap between the speed of AI development and the speed of government decision-making.

America therefore faces two risks at the same time. Moving too slowly could allow China and other rivals to gain a military and cyber advantage, while moving too quickly without safeguards could create powerful AI systems that are difficult to control. This tension is now at the center of the US national security debate.

The biggest question for America is no longer whether AI will become a national security weapon. That transformation is already happening across the Pentagon, intelligence agencies and defense industry. The New York Times' reporting makes clear that the harder question is whether the US government can build a consistent strategy before America's AI advantage disappears or the technology becomes too powerful to control.