Will AI be the new atomic bomb, a humanity tragedy?

With the advent of large language models (LLMs), the transformer architecture developed by Google’s AI team has given rise to the most profound achievement in rigorous application the world has ever known, from image recognition to chatbots. Now, with advancements in model architecture, GPUs, and software engineering, the scope of its applications is wider than before.

The war.

Back in the day, the U.S. showed its might in the advancement of military power in the 1991 Gulf War with Kuwait and Iraq. At that time, the superiority of precision-guided weapons allowed the U.S. to hit targets with certainty, which, in part, helped it win the war. Tactics are important, but so is the technology that enables them.

Now, when the news of Sam Altman signing his contract with the Pentagon is published, it officially opens the curtain of the new era of the war. The Agentic AI system that can operate 24/7 can monitor, analyze, target, and execute autonomously. Thus, the nation that can train the most capable and intelligent AI will likely possess the power to narrate global politics and influence others. As in the latest war, the Cold War between the U.S. and USSR, both countries were in a rivalry in multiple aspects, which was fueled by the monopoly of the atomic bomb. Truman’s administration asserted its power and influence over global politics. The same might or might not be true for Trump’s administration. According to the Economist, 90% of the AI frontier resides in the U.S. (GPT, Gemini, & Claude) and China (Qwen, Ernie, & DeepSeek). If this situation does not echo the atomic bomb era of the Cold War, then we are choosing to ignore what history has taught us.

So, what did history teach us?

I can’t think of anyone else, the American Prometheus, who gave mankind the greatest weapon and war. Oppenheimer would probably see this AI as similar to his atomic bomb, and perhaps expect it to give him the power to change the structure of world politics. In his time, as a salesman and scientist, he believed he had full control over his baby. He soon learned, after Hiroshima and Nagasaki, that he, a scientist, had no influence on the future use of the bomb. He tried so hard, lobbying and publicly advocating for his treaty on how to prevent the national use of the atomic weapon, which he failed to achieve. But his world of atomic peace is worth a look. He proposes that it will be a renunciation of sovereignty, where the atomic bomb is on the international agenda, no country will ever possess this weapon, and only scientists are allowed to use this atomic knowledge for the greater good. Of course, Truman did the opposite.

This time, in the AI era, things were different. The situation closely resembles what Oppenheimer had imagined: the development of AI is in the hands of private corporations, and some of the most advanced models are open-source, meaning that knowledge is shared among top scientists working on AI. That makes AI harder to contain and, as a double-edged sword, harder to govern. And make global politics even more complex. The question shifted from how AI can shape lives, organizations, and the world to whether humanity can govern it before its power destroys us.