The AI race is a data race. Africa cannot afford to watch.

The most powerful companies in the world right now are not the ones sitting on the largest oil reserves. They are the ones sitting on the largest reserves of human data. The AI race everyone is talking about is, at its core, a race for data. And the big tech firms are winning it because they already own the pipes, the platforms, and the people’s attention.

When AI Goes to War

In March 2026, reports confirmed that the U.S. military used Anthropic’s Claude AI during its strikes on Iran. According to NBC News, AI tools were being used to “sift through vast amounts of data” to identify and prioritise targets at a speed human planners cannot match. Claude was embedded in Palantir’s Maven Smart System, processing satellite imagery, signals intelligence, and intercepts in real time. The Washington Post reported that the U.S. managed to strike over 1,000 targets in the first 24 hours, partly because AI could build and rank scenarios faster than any room full of analysts.

What made the story even more revealing was the squabble around it. Anthropic tried to draw red lines to prevent mass surveillance or fully autonomous weapons. The Pentagon refused, wanting access to Claude for a rather vague “all lawful purposes”. When Anthropic pushed back, President Trump ordered federal agencies to drop Claude entirely.

Surveillance in Plain Sight

Still, you do not need to be in “enemy territory” to find yourself under surveillance. You just need to walk past someone wearing a pair of smart glasses. In October 2024, two Harvard students demonstrated exactly this. They built a system called I-XRAY using Meta’s Ray-Ban glasses, a facial recognition engine, and publicly available databases. Now, one could identify strangers on the street and pull up their names, phone numbers, addresses, and more, in seconds, without the subjects having any idea they were being scanned.

This was not a hypothetical. Multiple real-world cases have been documented: a man filmed women without consent at a university campus. Workers subcontracted by Meta in Kenya reported having access to sensitive and intimate data – from sexual activities to bank card details – for use in review and Meta AI training. Yet, over seven million units of Meta’s smart glasses were sold in 2025 alone.

Meta collected the data from these interactions. The AI got smarter. The users had no say.

Then there is WhatsApp. Meta AI is now embedded in WhatsApp and WhatsApp Business by default, rolled out globally without users opting in. Meta frames it as an “optional service” because you choose whether to engage with it. But turning it on was never your choice, and there is no toggle to turn it off.

For WhatsApp Business specifically, Business AI is enabled by default across all customer chats, including chats with regular WhatsApp users who have no idea they are now in an AI-mediated conversation. Those customers receive a small disclaimer the first time, but the interaction data flows to Meta either way. Turning it off requires actively locating the Business AI settings and disabling it manually. Most business owners never do. And any messages exchanged while Business AI is active are collected by Meta to improve its models, even after you disconnect the feature. The implications arrived quietly. The headline did not follow.

Data is Why Tech Giants Lead AI Development

None of this is accidental. Big tech leads in AI because big tech leads in data. Years of running the platforms where people socialise, shop, search, and communicate has given these companies a structural advantage that no amount of research funding alone can replicate. Training a large language model well requires not just compute but the right data, diverse, current, behaviorally rich data. Meta has it. Google has it. Microsoft has it. Anthropic has it.

The AI race is a data race. And the data was collected, often without meaningful consent, through the steady normalisation of surveillance as convenience.

What This Means for Africa

Africa has to confront some uncomfortable realities. The continent’s digital life runs almost entirely on foreign infrastructure. Social networks, cloud hosting, payment rails, messaging apps. As of mid-2025, Africa had just 223 data centres across 38 countries with foreign tech firms dominating the landscape, often shaping the terms of investment. Moreover, much of Africa’s critical national data, health records, financial data, identity systems, sits on servers owned and governed by foreign entities. If those systems go down, get compromised, or become politically weaponised, African governments have limited recourse. The recent hack of 150 GB of Mexican government data, including significant amounts of sensitive citizen data using Claude AI is a case in point.

What Needs to Happen

1. For policymakers:

• The work starts with strong, enforceable data protection laws; and African countries have largely risen to the challenge with 44 out of 55 African countries having passed data protection laws.
• The next step is enforcing data localisation requirements, not as protectionism but as minimum standards of sovereignty.
• Regional coordination under the African Union is also essential. Fragmented national policies will continue to give tech companies the room to forum-shop.
• Regulators also need to scrutinise the terms on which foreign AI tools enter public sector use, especially in health, policing, and finance, where algorithmic errors carry severe human cost.

2. For private businesses/actors:

• Know where your data lives. Cloud dependency is understandable, but businesses should understand which foreign jurisdictions govern their data, and what those governments can access.
• Invest in privacy-by-design principles early. When deploying AI tools, demand transparency on training data and model behaviour.
• Support local and continental tech infrastructure with procurement choices where viable.

3. For individuals:

• While the principle of data minimisation is thought to be upheld solely by data controllers/processors; you can decide to reclaim it and apply it in your daily life.
• Start with your settings. Most apps are set to maximum data collection by default. Change that.
• Be honest with yourself about which apps are truly necessary, especially those that ask for access to your location, contacts, microphone, and health data. Uninstall the unnecessary apps and delete the data they already have.
• Ask who benefits when you use a platform. Ask where your data goes and how long it stays. You will not always get a clear answer. But the habit of asking is itself a form of resistance.

Conclusion

The AI era is not coming: it is here. The question for Africa is whether the continent will enter it as a participant with leverage, or as a source of raw material for someone else’s intelligence. The difference will be determined by the choices made now, by governments, businesses, and ordinary people, before the architecture of dependency becomes too expensive to dismantle.

Written by Davida Opara; Communications Lead at Africa Privacy Roundup. Follow Africa Privacy Roundup for more insights on Africa’s digital governance evolution.