AI, data centres and digital colonialism: How intelligence became a commodity
&imwidth=600&imheight=450&format=webp&quality=medium "AI, data centres and digital colonialism: How intelligence became a commodity")
This photograph shows the logo of the US artificial intelligence safety and research company Anthropic displayed on a smartphone's screen in Brussels on June 10, 2026. Photograph: (AFP)
As governments impose AI export controls and tech firms build powerful models from public data, a debate is growing over ownership, access and digital power.
The most powerful AI created till now, Claude Fable and Claude Mythos made unavailable after a US government export control directive prohibited their use by any foreign national, whether inside or outside the United States. The directive cited that the model was susceptible to “jailbreak”. Anthropic's newly developed frontier “AI models” have been reported to have reached a level of capability that poses a catastrophic risk to public and national security. The US government has started to consider AI as infrastructure as a national security asset comparable to nuclear technology or satellite systems. It had earlier passed a regulation advising a 30-day "voluntary" review before making any AI models public. A discussion has been going on between the Trump administration and major AI companies for the government to effectively become a partner of the companies. This means countries like Russia and China would perceive AI as an extension of the US military.
AI models built on public data
According to Anthropic's System Cards, Claude Fable 5 and Mythos 5 were trained using publicly available information from the internet, harvested via their proprietary web crawler, ClaudeBot. Though the training process is more complex than just open data sets, these models are ingested with open forums, Wikipedia, research papers, and code repositories. Further, it also acquires datasets under commercial agreements with third-party businesses, data that its users or crowd workers provide, and data that people allow it to use, such as chats and coding sessions, to improve Claude.
So, while proprietary partnerships exist among the data mixes, they contain only a fraction of the trillions of tokens required to train an AI model.
Anthropic recently settled a class action lawsuit in 2025. The Bartz v. Anthropic copyright settlement involved a $1.5 billion payout regarding the ingestion of pirated books for AI training. The core irony is that while the building blocks of these models are mostly public, the resulting capabilities have triggered national security crackdowns, such as export controls.
Trending Stories
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
&im=FitAndFill=(700,400))
)
)
)
)
&im=FitAndFill=(700,400))
)
)
)
)
)
)
)
)
)
)
Intelligence as a commodity
OpenAI CEO Sam Altman said that they envision a future where knowledge is a commodity. By scraping the collective knowledge, culture, and open-source contributions of humanity, private corporations have built a new utility: digital intelligence. Now, when the government place security walls, and the corporations hide them behind expensive tokens, it is nothing but gatekeeping. Knowledge that was once free and decentralised, available for everyone's consumption, now becomes a proprietary black box; it is packaged, controlled and sold. Debashish Chakrabarti, Editor in Chief of the International Journal of Politics, argues that from biometric databases to cloud infrastructures, the Global South is generating the data that powers artificial intelligence — yet the wealth, control, and computational sovereignty remain elsewhere. To access that synthesised knowledge efficiently, users must now pay a toll to a private company. A kind of neo-feudal society, while the government will leverage that knowledge, use it as a geopolitical gatekeeping tool and create a class of knowledge elites.
Jai Asundi, Executive Director at the Centre for the Study of Science, Technology and Policy, thinks that export control is unlikely to be a permanent phenomenon. “Open-source models and tools will eventually catch up, and collective work by the global community is likely to make them as robust as—if not more robust than—proprietary systems. Humanity has benefited from globalisation, so while this ‘decline’ might be a temporary blip, it's unlikely to remain so over the long term,” he said.
But the extraction is not limited to mere gatekeeping. Massive data centre infrastructure is being built, which drains natural resources like water and land, and puts those at the lower strata of society at risk. For example, New Delhi had been aggressively courting US big tech to set up data centres in the country. The Indian government announced a 20-year tax holiday for foreign cloud service companies using India-based data centres to serve global customers. The aim is to position India as a global hub for artificial intelligence infrastructure. This push has collided with a severe reality on the ground. Farmers are being pushed to give up their land, as the protest continues. Similar pushbacks have been seen in places like Virginia, Iowa, and Arizona. Environmental groups are increasingly vocal about the carbon and water intensity of AI infrastructure. The basic structure has huge similarities with the colonial drain of wealth, where resources like power, water, and land are extracted from India, value is processed, and the surplus is flown back to wealthy foreign consumers. Indian consumers then pay for the digitised output as a service. The only difference between the colonial era and now is that India has agency, and it is voluntarily entering into this arrangement.
