In an age where Artificial Intelligence (AI) is rapidly transforming every aspect of life, the misuse of AI is becoming a growing threat particularly in matters of national security. A recent case example is involving an AI-generated deepfake of U.S. Secretary of State Marco Rubio. This incident highlights how sophisticated and convincing AI voice scams have become capable of targeting high-level officials. This deception shows a broader concern i.e. the weaponization of AI in geopolitics where misinformation, biased data, and algorithmic manipulation can trigger unintended consequences globally and regionally – especially in volatile nuclear dyads like India and Pakistan. As AI-driven technologies are without regulation, the risks of miscalculation, escalation, and even nuclear confrontation grow more acute. It demands urgent international cooperation and control mechanisms before catastrophe replaces competition.
Study shows that AI is the technological tool that helps machines learn from data. This eventually helps AI to decide smartly by recognizing patterns and predicting results. However, AI also has some limitations. First, its reliability depends on the quality and quantity of available data. Second, AI models are difficult to understand, as it is complex to explain how they process the data and come to a decision.
Additionally, AI’s decision-making is susceptible to distortion based on available data. Data holds bias according to national interests of different states. That is how algorithms make decisions that can seem unfair and skewed with sometimes even dangerous results. Imagine this playing out on a global scale, where nations rely on AI for intelligence, strategy, or even war. What happens when flawed AI reinforces bad assumptions i.e. misinformation? It might push leaders toward choices they wouldn’t otherwise make. That’s not just a hypothetical scenario – it’s a ticking time bomb.
Misinformation and deepfakes are not new, but AI turns it into something faster, sharper and harder to untangle, especially during a crisis where there is no room for a mistake. For instance, imagine two nuclear armed rival states, each of which is watching the other through a fog of AI-generated half-truths in a conflict. During the recent conflict between India and Pakistan from May 7-10, Indian media, on the night of May 8, Indian media began spreading propaganda claiming the destruction of Pakistani military assets, cities, and the capture of soldiers. By morning, the Indian public realized it was a widespread national disinformation campaign, which drew criticism from international and intellectuals, academicians, and ordinary citizens. This could have resulted in AI-based retaliatory responses during such a high-stake conflict. Such incidents highlight the risks of feeding biased or false data into AI systems, which, in future crises, could lead to miscalculations or escalations between nuclear-armed states. If it was a U.S.–China conflict, biased big data feeding AI systems could accelerate misperception and strategic miscalculation, with global economic and military consequences.
Further in 2019, India launched airstrikes in Balakot against Pakistan after a terrorist attack on an Indian military convoy in Pulwama in Indian Occupied Jammu and Kashmir. Now, picture an AI-fueled information enough to trigger a war – fabricated battlefield reports and deepfake videos showing destruction that never happened; social media algorithms – amplifying outrage where diplomacy is a lost cause. Such a conflict does not need more weapons. It needs one well-placed lie – the weaponization of AI.
In the meanwhile, competition on global level is inevitable between the U.S. and China if the weaponization of AI is not regulated or controlled by international authorities. Recently, a milestone was achieved by China where it announced a comprehensive regulation to avoid AI-generated disinformation. According to Chinese regulators, all AI-generated content must be labelled on the internet from September 1, 2025. Nevertheless, some threats are too big to handle alone. There is a need for rigorous efforts by both the U.S. and China as the world where AI controls civil and military decisions without safeguards is not just dangerous, it’s suicidal.
Historically, nuclear arms control came after the realization of both sides i.e. the U.S. and USSR along with scientists, politicians and strategists. Currently, arms control regarding AI is history repeating itself. Albert Einstein once warned that technology is racing ahead of human wisdom, dragging us toward catastrophe. Henry Kissinger identified these risks which later came to be known as “Kissinger’s Specter” – the fear that AI would not just change warfare but change us, and blur the line between reality and illusion.
In conclusion, deepfakes and AI-generated misinformation are accelerating the risk of strategic miscalculation globally, especially in nuclear-armed regions like South Asia. As AI-systems learn from biased data and their decisions can be flawed, unpredictable, and dangerous. Recent incidents such as fabricated war reports during the India–Pakistan conflict show how AI can escalate crises. Bearing in mind the above-mentioned case in the India-Pakistan nuclear dyad, there is a need to resort AI including countering fake and biased data. Urgent bilateral and international cooperation is essential. However, if the global AI arms race goes unchecked and spirals, there will be no room left for talking about competition, only survival after nuclear use. Many may not even survive, as there are no winners in a nuclear Armageddon!
Muhammad Ali Baig is a researcher at the Center for Strategic Studies (CISS), Islamabad. His X @alibaig111
