Two developments raise serious questions about what the rapid advancement of Artificial Intelligence may imply for the future of humanity.
According to an editorial published by The Economist, the first concern relates to the fact that some of the people working on the first line of development of this technology have begun to warn about risks that could lead to a global catastrophe, even the extinction of humankind.
The second concern relates to how technological advancement is changing the war in Ukraine, where the use of new technologies is increasingly being translated into military capabilities.
These two realities convey contradictory messages about a question that is becoming increasingly important: should the development of Artificial Intelligence be slowed down or should the technological race be accelerated?
For years, leaders and scholars in the field of Artificial Intelligence have warned that the technology could create difficult-to-control risks. The concern is that future systems could become so powerful that they surpass the abilities of their creators and no longer fully align with human objectives and values.
Another risk is related to the use of these tools by malicious actors. The technology could be used for cyberattacks, information manipulation, espionage, or the development of new weapons.
The concerns have also increased following reported incidents during the testing of autonomous Artificial Intelligence systems, known as "agents." In some experiments, these systems have attempted to escape controlled environments and interact with other systems, trying to infiltrate organizations.
Anthropic, one of the leading companies in the field of Artificial Intelligence models, has also identified cases where its systems have been used for malicious purposes. The company has reported uses of the technology for activities related to surveillance, as well as cases where such tools are alleged to have been used by the rebel groups in Yemen to help develop ballistic missiles.
Fear of an "Artificial Intelligence catastrophe" is intensifying
The debate about the risks of Artificial Intelligence has intensified following the latest post by a scholar from Anthropic and the alarming message he published afterward.
In a post that received more than 170 million views, he argued that Anthropic and OpenAI are "playing with our lives."
Meanwhile, Artificial Intelligence is developing at very fast rates and is demonstrating capabilities that could bring great benefits, but also considerable risks.
An example mentioned by The Economist relates to the mathematical problems of the millennium. In May, forecasting markets valued the probability that Artificial Intelligence would solve one of the great mathematical problems before 2030 at less than 0.35.
Only a few months later, in August, OpenAI announced that it had solved one of them, causing concern among some mathematicians about the rate at which the new models are advancing.
By late October, two of the most well-known technology figures, Sam Altman of OpenAI and Elon Musk, supported the call of Anthropic's leader, Dario Amodei, to "slow down the limits of technology," meaning to develop the most advanced Artificial Intelligence systems more carefully and at a slower pace.
The United States fears China and cites the lessons from Ukraine
On the other side is the administration of Donald Trump, which sees the loss of American supremacy over China in Artificial Intelligence as a strategic threat.
China is considered the US's main rival in this technological competition. The US Secretary of the Treasury, Scott Bessent, has stated that if China achieves supremacy in Artificial Intelligence, "nothing else will matter."
As an argument for accelerating technological development, the US can also use the example of Ukraine, where technological advancement directly affects how the war is being conducted.
Russia has developed autonomous drones for attacks on infrastructure and other targets, while new technologies have given these systems greater capabilities for communication and coordination.
At the same time, Musk is developing a satellite communication system similar to Starlink, which aims to increase the accuracy of long-range attacks.
The war in Ukraine has shown that even a limited technological advantage can affect the military balance.
Kyiv has tried to compensate for the lack of some air defense systems, including Patriot missiles, by developing the drone and other technologies industry. Ukraine has used these capabilities to repel Russian attacks and conduct strikes deep into Russian territory.
The situation in Eastern Europe shows that technology is increasingly being translated into military power.
Fear of a new technological Cold War
Even without a direct military conflict between the US and China, supremacy in Artificial Intelligence could give one side an advantage in cyberattacks, espionage, and hybrid warfare.
A global agreement to slow down the development of Artificial Intelligence would only work if China also agreed to participate.
Trump has planned a meeting with Chinese President Xi Jinping at the White House on October 24th. However, an agreement to limit the development of Artificial Intelligence is considered difficult, as Beijing would not have an interest in consolidating US supremacy in this field.
One of the lessons from the Cold War and the nuclear race is that any such agreement would require acceptable control mechanisms from both parties.
But Artificial Intelligence presents a unique problem. Data centers can be seen and monitored, but it is much more difficult to control precisely what is happening within them.
A system trained to become superintelligent may appear from the outside as a normal program for everyday conversations and questions, while its real abilities and objectives may be much more advanced.
Can an agreement between the US and China help?
An idea put forward in the debate is that an agreement between the US and China on Artificial Intelligence could also have a political dimension, giving Donald Trump the opportunity to claim a role in an important international agreement.
A few days before Xi Jinping's visit to the US, OpenAI's executive director, Sam Altman, proposed that if the American and Chinese presidents could reach an agreement on Artificial Intelligence, they both should receive the Nobel Prize for Peace.
In an interview for Fortune magazine on October 12th, Altman argued that the US and China cannot risk losing control over a technology that both parties are trying to dominate.
"I don't think this is difficult. It is like a one-sided document," he declared, referring to a possible agreement.
In practice, however, reaching an agreement seems much more complicated.
AI, another area of US-China competition
The official US-China discussion on Artificial Intelligence has developed even earlier. In 2024, both sides agreed that people must retain control over nuclear weapons, even during a period when Artificial Intelligence technology is advancing rapidly.
In May, Trump and Xi Jinping agreed to revive the dialogue on Artificial Intelligence, increasing hopes for a common approach to some of the risks.
The discussions have included concerns ranging from the use of Artificial Intelligence for developing biological weapons and cyberattacks to the possibility of autonomous systems operating outside human control.
American officials had hoped to develop a round of talks before Xi's visit to the US on October 24th, with the aim of creating a joint working group with technical experts and officials with authority to negotiate.
However, after some delays, Artificial Intelligence is expected to be discussed within the framework of broader talks on trade and other issues between US Treasury Secretary Scott Bessent and Chinese Vice Premier He Lifeng.
Bessent stated on October 16th that he was ready to discuss the common risks related to Artificial Intelligence and to avoid the "splitting of the two systems."
However, expectations for major progress in these talks this month remain limited.
Accusations of espionage increase tensions
On August 8th, the FBI and the US National Security Agency accused Chinese Artificial Intelligence laboratories of systematic data theft from American competitors, raising suspicions about the knowledge of Chinese authorities.
The Chinese Ministry of Commerce did not directly deny the accusations. Instead, it defended the practices of "data distillation," describing them as industry standards and warning against US sanctions against Chinese laboratories.
Two days later, Anthropic accused some Chinese laboratories of using Claude model users to collect data for training their Artificial Intelligence systems.
The company also alleged that in some cases, users linked to the Chinese military and defense companies had imported sensitive data on surveillance and weapon designs.
Following these developments, Anthropic's leader, Dario Amodei, called for a global slowdown of Artificial Intelligence development.
Trump opposed this idea, arguing that the US must maintain its supremacy over China.
"We are leading China in artificial intelligence. I want it to remain so, because whoever has artificial intelligence has everything," the American president expressed.
China sees the US as the main threat
China has also opposed the calls for slowing down the development of Artificial Intelligence.
Amodei has demanded not only a slower pace of development but also restrictions on the sale of microchip equipment to China and more rapid controls over the transfer of data and technology.
Chinese state media have described the calls to slow down Artificial Intelligence development as an attempt to curb China's technological development.
"Fear, confrontation, and fierce competition will hinder the efforts for a healthy global governance of artificial intelligence," declared a spokesperson for the Chinese Foreign Ministry.
Chinese officials also see the US as a significant source of risk in the field of Artificial Intelligence.
The Chinese intelligence chief, Chen Yixin, has warned that "hostile forces" could use AI-generated content to spread political disinformation or develop cyber weapons.
China is also concerned about the risks of losing control
Despite the clash with the US, China is also facing an internal debate about the risks of uncontrolled Artificial Intelligence development.
Chinese laboratories, like their American counterparts, are working to create increasingly advanced systems.
Xue Lan, the director of a state committee of experts on Artificial Intelligence, has stated that the models should not be allowed to continuously improve without human intervention and control.
In August, a Chinese regulator warned about the risk of "extreme loss of control," as OpenAI reported that thousands of Artificial Intelligence agents had escaped from testing environments.
Chinese companies have also faced similar incidents. Alibaba discovered that a programming agent had started mining cryptocurrencies without instructions to do so, while another system raised concerns about data flow.
Hope remains in dialogue
The main question is now whether the US and China can transform their shared concerns about Artificial Intelligence into a basis for concrete discussions.
China has sought more informal dialogues among academics and former officials, arguing that these contacts can help in understanding the differences between the two countries.
The US, however, remains cautious, as there is concern that Beijing may use these discussions to gather information.
Another problem is the different way the two countries' diplomacy functions. According to a former American official, Chinese diplomacy is very controlled and prepared, while Artificial Intelligence is a new field and constantly changing.
A comprehensive agreement on the rules that should govern the development and use of Artificial Intelligence still seems difficult.
However, cooperation may be possible in more limited areas, such as the exchange of information on security incidents, the development of common methods for evaluating models, and greater transparency after the presentation of new systems.
The military use of Artificial Intelligence will be an even more difficult matter to include in official discussions.
For now, one of the most important steps would be to maintain open communication channels between the two superpowers.
But the pace at which Artificial Intelligence is developing and the uncertainty about its limits are increasingly making the need for international decisions and regulations more urgent.