Concerns about the potential dangers of artificial intelligence (AI) have intensified in recent days, following a series of warnings from leading figures in the technology sector. Jacob Coxon, a 27-year-old researcher at AI firm Anthropic, resigned last week, publicly urging caution by stating that AI developers are “gambling with our lives” and suggesting the technology could pose an existential risk within a decade.
His departure was soon followed by an open letter from Anthropic’s CEO, Dario Amodei, who cautioned that within six months to a year, autonomous AI agents capable of self-improvement could take control of the internet, potentially causing hundreds of billions of dollars in damage. Amodei called for a slower development pace among AI companies to ensure proper alignment and safeguards are in place. He also urged the United States to work with China to mitigate acute risks associated with the technology. The letter received endorsements from notable figures including Elon Musk and Sam Altman, CEO of OpenAI.
The warnings have sparked debate among prominent voices in the field. Former U.S. President Donald Trump dismissed the safety concerns as a “hoax” on social media, emphasizing his belief that a strong and intelligent presidency, such as his own, provides sufficient oversight. Trump criticized Amodei for what he described as posturing and argued that stringent AI regulations could undermine America’s competitive edge, particularly in relation to China. Nvidia’s CEO Jensen Huang also rejected Coxon’s warnings, labeling them as “not grounded in science.”
This debate highlights two central and unresolved questions: whether AI is principally a disruptive force or a destructive one, and if destructive, what form that destruction might take. Historically, technological innovations have caused significant disruptions in labor markets but have ultimately led to job creation and economic growth. A recent Stanford study found that firms adopting enterprise AI experienced a 10 percent increase in employment over two years, challenging fears of mass job displacement.
However, some experts caution that AI may represent a fundamentally different kind of risk—one comparable to the development of nuclear technology or controversial scientific research with unforeseen consequences. Examples from history include the 2014 moratorium on gain-of-function research under President Barack Obama, intended to study the risks of creating potentially dangerous viruses, and President Harry Truman’s establishment of the Atomic Energy Commission in 1946, which provided a civilian framework to regulate nuclear technology.
These historical precedents suggest that while slowing down risky research might be morally appealing, it may prove ineffective if other global actors do not follow suit. Moreover, unlike nuclear technology, AI tools and knowledge have never been fully contained and have proliferated widely since their inception, making attempts to regulate or constrain development more complicated.
Given these challenges, some experts argue that accelerating research into AI safety and control mechanisms may be the most practical course of action to address potential threats. Although AI presents unprecedented ethical and existential questions, past experiences with transformative technologies highlight the need for caution, international cooperation, and robust governance structures to navigate the uncertainties ahead.