The global conversation around artificial intelligence is broken. Western leaders spend their time debating hypothetical doomsday scenarios while tech billionaires rush to build increasingly massive server farms. Meanwhile, the actual, real-world damage of unregulated software hits ordinary people every day.
When world leaders gathered in Evian, France, for the G7 summit, the traditional talk of trade barriers and military alliances took a back seat to a more immediate threat. Indian Prime Minister Narendra Modi stood before the assembly and shifted the focus entirely. He didn't focus on sci-fi horror stories about rogue robots. He focused on what happens when humans become nothing more than raw data points for corporate algorithms.
The core issue isn't whether machines will become smarter than us. The issue is how we protect human dignity, prevent the exploitation of children, and stop deepfakes from tearing apart the fabric of open societies right now.
Moving Beyond Machine Sovereignty to the Manav Doctrine
Most global policy discussions treat AI like a runaway train that we can only hope to steer slightly. This approach is completely wrong. At the G7 outreach session titled Ensuring a Safe, Rapid and Efficient Rollout of AI, India introduced a completely different framework called the MANAV vision. It literally translates to "human" in Hindi, and it breaks down into a clear, pragmatic philosophy.
- Moral and ethical systems that govern machine behavior from the first line of code.
- Accountable governance so companies can't hide behind automated decisions.
- National sovereignty over data ownership, preventing digital colonialism.
- Accessible and inclusive tools that don't just benefit the wealthiest tech hubs.
- Valid and legitimate outputs that users can actually trust.
This isn't about halting innovation. It's about maintaining human control. Think of it like a GPS system. The satellite maps out the entire city and suggests the fastest routes, but you are still the one holding the steering wheel. You choose where to turn. Current tech development wants to put the car on autopilot without asking if you even want to go to that destination.
The true test of any software isn't how many billions of parameters it can process in a millisecond. It's whether it makes life better, safer, and more dignified for an ordinary person who doesn't know a thing about coding.
The Exploitation Playground and the Threat to Children
The most chilling part of the G7 address focused on how unregulated software affects children. AI has the potential to customize education, translate complex ideas instantly into regional languages, and spark incredible creativity. But right now, we are failing to protect young people from the dark side of that same technology.
Without strict, built-in governance, the digital space turns from a learning environment into a massive tool for manipulation. Children are currently exposed to targeted misinformation, algorithmic radicalization, and AI-generated exploitation material.
French President Emmanuel Macron directly addressed this during the summit, appealing to global leaders to restrict unsupervised digital access for children under 15. The consensus is clear: just as we carefully curate and vet school textbooks to ensure they are safe and age-appropriate, we have a duty to ensure the AI space is family-guided and fundamentally secure.
Digital Watermarks and the Battle Against Fabricated Reality
Deepfakes are no longer just a minor internet nuisance or a tool for funny viral videos. They are actively destabilizing democratic institutions and destroying personal reputations. We saw this clearly during recent political cycles, where synthetic audio and morphed videos were deployed systematically to manipulate voters.
The solution isn't to simply tell people to "be more careful" online. That puts the entire burden on the victim. We need to enforce a policy where safety is built into the design phase of the product.
The Case for Mandatory Authentication Labels
We need to treat digital content the same way we treat food or medicine. You wouldn't buy a bottle of pills without a verified safety seal, and you shouldn't consume media without knowing exactly where it came from.
- Cryptographic Watermarking: Every AI model must embed invisible, unalterable digital signatures directly into the metadata of generated text, images, and video.
- Origin Tracking: Platforms must display clear provenance labels so users can see the exact history of a file—whether it came from a real camera or an algorithm.
- Regulatory Sandboxes: Governments need to create controlled testing environments where new models are evaluated for safety, bias, and deepfake generation capabilities before they are allowed to launch publicly.
If a company wants to deploy a commercial flight, they have to pass rigorous aviation safety checks. If a shipping company wants to launch a cargo vessel, they must comply with strict maritime laws. Yet, tech companies regularly launch incredibly powerful, highly disruptive software models to millions of users with zero prior safety testing. That dynamic has to change.
Preventing the Great Split Between North and South
If we leave AI development entirely to a small handful of trillion-dollar corporations and wealthy nations, we will create a permanent economic divide. United Nations Secretary-General Antonio Guterres noted at the periphery of these talks that the future of human intelligence cannot be left to the personal whims of a few billionaires.
The Global South cannot simply be viewed as a massive pool of cheap data to train Western models, or as a passive market to dump finished software products. Cyberspace is a global public good. If democratic nations want to protect their critical infrastructure and counter sophisticated cyber threats, they need open, equitable access to foundational AI models.
True democratization means building tech infrastructure where it is needed most. This requires massive investments in local data centers and physical hardware. For instance, India's recent push to procure 50,000 graphics processing units (GPUs) within a short six-month window shows what it looks like when a nation takes its digital sovereignty seriously. It allows a country to build its own localized models rather than relying entirely on foreign tech monopolies.
Practical Steps for Business Leaders and Policymakers
We cannot wait for a perfect global treaty to start fixing these issues. Organizations and local governments need to take immediate, practical steps to protect their data, their employees, and their constituents.
First, stop treating AI implementation as a purely technical issue managed by your IT department. It is a core governance issue. Every business using automated tools must audit their pipeline to ensure they aren't accidentally feeding proprietary company data or sensitive customer information into public models.
Second, demand transparency from your software vendors. If you are purchasing automated HR tools, customer service bots, or content generators, demand to see their safety testing documentation. Ask them exactly how they prevent deepfakes, how they watermark their outputs, and where their training data comes from.
Finally, invest heavily in internal digital literacy. The single best defense against corporate espionage and deepfake phishing scams is a workforce that knows how to spot synthetic media. Train your team to look for the telltale signs of audio manipulation, verify unexpected requests through out-of-band communication channels, and treat unverified digital media with healthy, deliberate skepticism.
The geopolitical discussions at the G7 summit made one thing incredibly clear: the era of giving tech companies a free pass to move fast and break things is officially over. The command of this technology must remain firmly in human hands.
To see a practical breakdown of how these synthetic media threats manifest in real-world scenarios and what experts are doing to counter them, check out this detailed analysis of Deepfake Politics and India's AI Race. It provides an excellent look at how these abstract policy debates translate into actual security challenges on the ground.
