Deploying Geopolitical Cards in the AI Era

As Artificial Intelligence (AI) picks up pace - scale, magnitude, frequency and the widespread use case innovations - it has also got governments nervous. They are now investing in developing strategies to navigate, manage and contain the new and unprecedented level of real risk that dual use AI technology ecosystem emanates.

It is now recognized that AI, once seen as a technological marvel and a bedrock of national power, security and economic dominance, has emerged as a powerful tool to mold public opinion and destabilize countries. Even big powers like the US and China stand exposed and vulnerable.

The International AI Safety 2026 Report points out that AI generated content can produce measurable changes in people’s beliefs. People interacting with content produced by models with huge computing power are more likely to change their views. Further the Report points out that AI generated content can be difficult to detect. What is alarming is that the “gap between pace of advancement and the ability to implement effective safeguards remains a critical challenge”.

AI now has the demonstrated capability to mount scalable cyber-attacks - be it public databases or super secure secret military data stores. Recent events around the globe have shown that AI is very adept at exploiting software vulnerabilities in firewalls and protection layers to gain unauthorized access. Two widely reported cyberattacks relate to Australia and the Pentagon.

Prime Minister Anthony Albanese of Australia recently announced that an OpenAI agent unlawfully accessed public and non-public Medicare data.

The Pentagon recently confirmed a breach of a Defense Manpower Data Center (DMDC) system that exposed data of over 2.76 million living people. The Pentagon, however, did not confirm who accessed the data.

Instances of ‘AI assisted malicious attack’ and other AI enabled threats abound. But these threats take on a different dimension when they are used by sovereign states as a coercive toolkit of choice.

AI generated content when amplified by social media platforms, repeated over networks and egged on pointed narratives can have differentiated impacts on demographic segments. Experience shows these are usually malicious.

Social unrest in recent years has all displayed a common pattern - a specific demographic segment’s injured sense of justice is expanded until it inflames large sections of society to finally unseating elected governments. Connecting the dots, it is not difficult to see the high level of coordination to manipulate public judgement that ends up with the changing the regime in the affected country.

In other words, a cognitive class divide has already taken deep roots in societies, thanks to AI assisted content that has been accelerated by social media. These manufactured chasms are tinder boxes waiting to inflame at the appropriate cue. Governments are fully aware that these are not fictional scenarios, but perilous situations that can crop up overnight and overthrow them.

The recent concurrent street protests by ‘students’ in Albania, France and India are strikingly similar in concept, coordination and execution. Local governments have reported the hidden hands of big networks with high degree of coordination and deep pockets in the unrest.

Of the 54 organizations and movements identified as part of the nationwide ‘students protest’ that led to 2,500 arrests across France over the weekend, only 16 are student or youth-led groups, according to a Fox News Digital report.

A combination of AI enabled cyber-attacks and influence engineering operations continue to succeed in exposing the soft underbellies of societies that could swiftly descend into chaos. They would be indispensable tools of statecraft.

Now, AI appears fully integrated as a core element of state sovereignty and instrument of state policy while simultaneously being integral to economic and national security. In this context it is easy to understand the clamor of countries to retain control of domestic compute, local language models, and AI infrastructure.

But the real winner(s) in the game will be those who can create LLMs at scale and control access to the AI ecosystem – infrastructure, data and usage.

Only those countries with the technical, financial and industrial ecosystem can build their own LLMs and appropriate AI enabled strategic response systems. Very few countries have affordability and access to such tools. This AI tiering will have far reaching geopolitical and geoeconomic consequences in the years to come.

Currently, only the US and China are in leading positions in the AI tiered intelligence economy. However, the EU, India along with other select smaller countries are working overtime to catch up.

India’s strategic initiatives include investments in the development of indigenous AI models (Sarvam AI, Gnani.ai, BharathGen) to limit dependence on external technologies.  Europe may not be far behind. As can be expected, this is evolving fast and a lot of changes are expected on this front.

Global initiatives to put in place suitable regulatory guardrails and governance architecture for AI are also afoot. But the geopolitical fracture on AI is evident. International AI summits have highlighted deep division between US and other nations with wherewithal to compete.

The Paris AI summit in February 2025 was an important milestone in chalking out a governance roadmap for AI. Signed by 58 nations (excluding US and UK), the summit declaration included general principles such as accessibility, overcoming the digital divide and developing AI that is open, transparent, ethical, safe, and trustworthy.

The Delhi AI Impact Summit in February 2026   succeeded in pivoting towards democratized access, diffusion, adoption and inclusive growth while integrating with Digital Public Infrastructure (DPI). Championing the cause of democratized access is seen as speaking up for an inclusive global South.

These developments underscore the seriousness of the geopolitical and geoeconomic risks that lurk for the world.

Many pundits have advocated an NPT (Nuclear Non-Proliferation Treaty) like treaty or agreement among nations on the safety and ethical use of AI. But this may face a lot of hurdles given that the NPT has split the world into nuclear haves and nuclear have-nots.  

What is plausible is self-regulated industry-led guidelines for the safe deployment of AI. But AI governance is evolving and the final shape and structure will be anybody’s guess.