How algorithms capture the youths minds

THE Jantar Mantar protests in Delhi should have been a textbook exercise in democratic expression; instead, they became a case study in algorithmic capture. The Delhi police identified over 180 social media handles spreading misinformation during the demonstrations, including AI deepfakes and manipulated content. They include a viral claim that a woman police officer had died, a fabricated post that the army had been deployed, and a synthetic image of a blood-covered hand holding a portrait of BR Ambedkar over a police barricade, entirely AI-generated, yet potent enough to inflame passions. Investigators traced more than 400 Pakistan-linked handles amplifying these narratives. But the manipulation cut deeper than foreign handles. Young protesters on the ground were fed algorithmic narratives that distorted the scope of their constitutional rights, weaponising the language of liberty to provoke confrontation with law enforcement, campaigns dressed in the vocabulary of "right to expression" convinced them that provocation was protected speech, defiance was duty and restraint was authoritarian overreach. Students who arrived to exercise legitimate democratic rights were manipulated into behaviour that undermined the very cause they believed they were defending. This is not misinformation. This is cognitive warfare and India's youth are on the front lines. The World Economic Forum's Global Risks Report 2026 placed mis- and dis-information among the top short-term global risks, noting it is one of the few risks that remains severe over both the two- and 10-year horizons and catalyses or worsens all other risks. The report documents how AI systems now optimise content for maximum emotional impact, using psychological profiling and emotional triggers to manipulate public perception and fuel polarisation. India is now the world's most exposed democracy to this threat. Forty-seven per cent of Indian adults have encountered AI voice-cloning or deepfake scams, nearly double the global average. Leading fact-checkers recorded that AI-generated content accounted for over 20% of all debunked material in 2025, more than double the previous year's figure. The average Indian smartphone user spends over four hours a day on platforms whose algorithms are explicitly designed to maximise engagement through emotional arousal. That is not a design flaw. It is the business model. The 2026 social media algorithm has evolved into something far more sophisticated than simple content recommendation. It now employs computer vision to identify objects and faces, semantic analysis to scan captions and audio transcripts for intent, sentiment detection to measure emotional tone and behavioural biometrics to track micro-actions like hover time, scroll speed and re-watches. Content is tested on seed groups of 100-500 users; if retention exceeds 70%, the algorithm identifies "lookalike interests" and surges the content to broader audiences. This means the most polarising, emotionally provocative content is systematically amplified because it triggers the engagement metrics the algorithm rewards. A young Indian scrolling through her feed is not seeing what is true or important. She is seeing what has been algorithmically proven to make her angry enough to share. When an algorithm maps your fears, targets your prejudices and feeds you synthetic realities calibrated to keep you scrolling, it does not merely misinform you. It colonises your cognition. And when that algorithm is deployed across 800 million internet users in 22 languages, with uneven digital literacy and deep social fault lines, you have the architecture of manufactured consent operating faster than any fact-checker can run. The government has acted, but the framework remains incomplete. In February 2026, MeitY notified amendments to the IT Rules regulating "synthetically generated information" (SGI), mandating prominent labelling of AI-generated content, permanent metadata for traceability and a three-hour takedown regime for unlawful synthetic material. Significant social media intermediaries must require users to declare AI-generated content before upload. Non-compliance exposes platforms to loss of safe harbour protection. This makes India one of the first major economies to statutorily regulate deepfakes and it is a significant shift from reactive moderation to proactive accountability. But takedowns are triage, not treatment. A three-hour removal window is meaningless when a deepfake has already reached 50 million WhatsApp forwards. Labelling is useful only if the user can read it, understand it and trust it; a questionable proposition in a country where digital literacy varies dramatically by region, gender and class. More critically, the amendment does not touch the deeper engine of harm: the recommendation algorithm itself. It regulates what content is created, not how cognition is manipulated. What is missing is a framework for cognitive protection. We need mandatory algorithmic transparency and not the black-box opacity that platforms currently enjoy. We need a prohibition on emotionally manipulative micro-targeting for political purposes. We need digital literacy mandates that treat cognitive defence as a fundamental skill, not an optional extra. Most importantly, we need a statutory AI algorithm review committee — a nationwide body with technical expertise, investigative powers and regional presence mandated to conduct regular audits of recommendation engines across major platforms. This committee must examine whether algorithms systematically amplify polarising content, particularly during election periods and national crises, and whether they suppress diverse viewpoints in favour of engagement-driven conformity. Such a body would not censor speech; it would expose the architecture of manipulation so that regulators and citizens alike can see how attention is being harvested. This is not merely about stopping harm. It is about preserving what makes us human. A nation of 1.4 billion cannot afford to become a population of passive consumers, fed pre-digested outrage by machines that know our vulnerabilities better than we know ourselves. We must build safeguards that keep human creativity and critical thinking at the centre of our digital life. Our education system must teach young Indians not just to code, but to question, to recognise when an algorithm is nudging them towards a conclusion rather than offering information. A recommendation engine that shows you what you already like is not a service. It is a cage. For India to grow, her people must remain capable of independent thought, dissent and imagination, qualities no algorithm can replicate and no democracy can survive without. The European Union is already enforcing algorithmic transparency through the Digital Services Act. The US is debating federal cognitive protection legislation. China regulates attention architecture through state control. India risks becoming the jurisdiction where 1.4 billion minds are treated as raw material for engagement metrics; a democracy where consent is not manufactured by argument, but engineered by algorithm. If Parliament does not move beyond takedowns to regulate the architecture of manipulation through an empowered algorithm review committee, mandatory audits and a national commitment to cognitive independence; India will face a predictable catastrophe: elections decided not by debate but by deepfakes, social cohesion shredded by synthetic outrage and a citizenry that no longer shares a common reality. The code is already writing our thoughts. If we do not write the law that governs it, we will discover that the only thing more dangerous than an unregulated AI is a population of 1.4 billion that no longer knows what it believes, no longer creates what it imagines and no longer questions what it is told. That is not the India we were promised. That is not the India we can afford to become. Aditi Ahlawat is Director, AI Agents, Mastercard