Recently, an increasing number of scientists working in artificial intelligence have been warning, quite plainly, that "we are all going to die" unless limits and restrictions are immediately placed on the current pace of AI progress.
Other analysts speak of exaggeration, implying that the risks are being inflated, even noting that alarmist predictions have accompanied technological revolutions as far back as the Industrial Revolution. There is no doubt that artificial intelligence is evolving into the new arena of major competition between the two superpowers, the US and China, much like the "arms race" or the "space race" between the US and the USSR once was.
However, the "AI Apocalypse" is not going to be stopped or slowed down by US President Donald Trump. And that is because he wants to achieve a major victory…
The Apocalypse
Jacob Coxon, a researcher who had worked at both OpenAI and Anthropic, argued that "we really seriously believe that artificial intelligence could kill all humans. Personally, I consider the probability to be over 10% within the next decade." As Coxon said, the hardest question one is called to answer is what an "AI Apocalypse" would actually look like.
One of the risks described by Dario Amodei, CEO of Anthropic, was the creation of a swarm of bots that would act as supercomputers capable of taking over the internet. Coxon noted that such a scenario could become reality within the next six months to a year.
In public statements, Sam Altman, head of OpenAI, said that the development of AI could lead to two particularly negative scenarios: the loss of control over the future to artificial intelligence and the concentration of excessive power in the hands of a single person or company.
Indifference
As CNN reports, if a terrorist threat led experts to warn of an immediate, existential threat to humanity, the entire mechanism of the US government would be mobilized in an immediate and unconstrained effort to prevent catastrophe. However, a remarkable sequence of warnings from leaders of artificial intelligence companies—according to which their creation may soon exceed human control capacity—has so far been met with near indifference by the Republicans controlling Washington.
President Donald Trump stated on Sunday that "negative forces" are inventing "things that are not going to happen." House Speaker Mike Johnson dismissed the call for emergency legislation, saying that AI executives "pushing the limits" could slow down if they wanted to, but that Congress should not rush to act.
Trump: We are ahead of China
Both also warned that slowing down the US artificial intelligence industry would give China an advantage in the race for superintelligence—thus framing the discussion less as a safety issue and more as a matter of national security. Indeed, over the weekend, the leaders of the United States and China outlined their visions for the technology that is expected to define the new era. Their approaches were at almost opposite ends of the spectrum.
Setting aside calls from AI scientists and executives from Anthropic, OpenAI, and Google to slow down development, US President Donald Trump stated that the most important thing is for the United States to maintain its lead over China in the increasingly intense superpower competition in AI. "We are ahead of China in AI... and, frankly, I want it to stay that way, because whoever wins in AI, wins," he said on Saturday.
China wants a more open model
On Sunday, Chinese leader Xi Jinping delivered his own call for more international cooperation. Countries should "encourage open source, transparency, cooperation, and sharing," he said in a speech in India to leaders of the Global South, in which he reiterated his proposal for a global AI governance framework—without, however, providing details on what such a framework would look like.
The juxtaposition of the messages from Trump and Xi highlights the lack of consensus and the deep trust deficit between the two superpowers in the high-stakes race for AI dominance.
Strategic advantage
The United States appears to be increasingly turning its dominance in this specific sector into a tool of strategic leverage, treating it as a strategic advantage to be protected and exported to friendly nations, while China, lagging behind in the competition, proposes a more open model.
AI is expected to dominate the agenda of the long-awaited Trump-Xi summit, which will take place in two weeks in Washington. In their May meeting in Beijing, the two leaders discussed safety guardrails for AI, but the results were limited, while last week the Chinese Ministry of Commerce announced that the two leaders agreed to establish an intergovernmental AI dialogue.
However, due to their broader geopolitical rivalry, experts have low expectations for any meaningful progress.
How big a threat is China?
Chinese AI has lagged behind its American competitors for years. This is partly due to US technological restrictions that limit Chinese companies' access to advanced AI processors—essential for training increasingly complex artificial intelligence models.
However, last year, Chinese AI startup DeepSeek shocked the global industry with a model that approached the performance of top American systems, using, according to its own claims, far fewer computational resources. Since then, Chinese AI labs have been evolving their models at a faster pace, with Alibaba, Z.AI, Minimax, and, more recently, Moonshot presenting equally capable models that challenge the dominance of leading American companies such as Anthropic and OpenAI.
Accusations of theft
Alongside the progress of Chinese models, however, American accusations of cheating have grown increasingly vocal. Washington, as well as OpenAI and Anthropic, have accused Chinese companies of stealing American intellectual property by training their models on the outputs of advanced US models, a process known in the industry as "distillation." Last week, US federal agencies, including the FBI, the National Security Agency, and the Cybersecurity and Infrastructure Security Agency, claimed that Chinese AI companies are committing "industrial-scale" theft against their American competitors.
Malicious actions
In an extensive report last week, Anthropic also detailed attempts by malicious actors, including alleged entities linked to the Chinese government, to abuse its systems to develop weapons and conduct surveillance on minorities, such as the Uyghurs in Xinjiang. In one example, Anthropic revealed sensitive information regarding Chinese surveillance attempts because queries from a user—believed to be connected to the Chinese military—sent to Moonshot's Kimi model were rerouted to Claude (editor's note: Claude is an artificial intelligence system by Anthropic similar to OpenAI's ChatGPT).
In this way, according to the report, Moonshot's secret effort to divert user requests to Claude was exposed. The report added that other Chinese companies, such as DeepSeek, had engaged in similar practices.
Beijing denies accusations
Beijing has rejected the charges, while Chinese AI labs have largely maintained silence. Moonshot declined to comment on the report to CNN last week, while DeepSeek did not respond to a request for comment.
"Promoting various threat narratives, pursuing confrontation, and engaging in malicious competition will only succeed in disrupting efforts toward global AI governance and will serve no one's interests," Foreign Ministry spokesman Guo Jiakun said on Monday. "All sides should work together to promote an AI that is open, inclusive, accessible, and beneficial to humanity," he added.
What does China propose?
Just as the United States feels increasingly threatened by rapidly evolving Chinese models, Beijing is equally concerned about the cyber capabilities of systems developed by US companies—capabilities that became evident this summer when agents from OpenAI and Anthropic acted uncontrollably and carried out unauthorized breaches of other systems. Beijing also perceives AI safety risks differently than the American perspective on existential risks, such as the possibility of AI agents acting out of control. In addition to prioritizing political control and maintaining social stability, it views the United States as a "hegemonic threat" to its own technology development.
"Cold War playbook"
In an article on Sunday, Minister of State Security Chen Yixin highlighted the dangers of "hegemonic practices," in a veiled swipe at the United States for "creating entity lists, imposing tech restrictions, monopolizing industry standards, and creating closed-source ecosystems" in the name of national security protection.
"These actions fragment the global AI industry and supply chains, exacerbating imbalances in the development of the global AI industry," he wrote.
An editorial in the state-backed newspaper Global Times over the weekend also rejected the recent essay by Anthropic CEO Dario Amodei. "On the surface, it appears to be a 'rational statement' focused on global AI safety," the editorial noted. "However, a careful reading reveals its hidden agenda: The article is filled with restriction provisions targeting China and, in essence, constitutes a 'Cold War playbook' for the AI sector."
Are they worried about AI in China?
While prominent executives and researchers in the United States have publicly voiced their fears regarding the risks of the technology, expressing similar concerns in China—where the ruling Communist Party strictly controls the media and monitors the population—is much more difficult and entails far greater risks. China's tech sector also operates within a much more centralized, top-down regulatory environment, where the government establishes extensive rules regarding what AI systems can produce and how they can be used. China's national AI safety standards, for instance, define requirements regarding training data, privacy, security, and prohibited content.
There is also greater public optimism about AI in China.
A 2025 Edelman survey found that 72% of respondents in China said they trust AI, compared to just 32% in the United States. And although Chinese AI labs have rapidly closed the gap with their American competitors, they still lag behind the most advanced US systems. This may give Chinese companies and researchers less incentive to voluntarily restrict the development and deployment of increasingly powerful AI systems as they try to catch up.
Can the two powers reach an agreement on AI?
There appears to be little reason for optimism. Despite the commercial truce achieved between the two countries in October of last year, tensions continued to flare up. The fundamental obstacle is trust. Washington fears that cooperation could give China access to technology that would enhance its military and economic power, while Beijing views "national security" as an American excuse to restrain China's technological rise.
A post last month by Yuyuantantian, a social media account affiliated with Chinese state broadcaster CCTV, outlined the terms for US-China dialogue on AI, stating that Washington must guarantee that United States AI companies will adhere to any agreed rules and regulations before negotiations can begin.
"A clear distinction must be made between real security threats and simple technological competition," it noted. "The real question is not whether there should be governance, but who should define the safety boundaries of these advanced models."
www.bankingnews.gr
Σχόλια αναγνωστών