Rajeev Narayan
India recently woke up one morning to discover that an AI-generated fake video of a prominent political leader had gone viral before sunrise, spread across crores of screens, weaponised by anonymous accounts and consumed as truth. The clip was fabricated. The outrage was real. People took to the streets, forcing traffic to a standstill. Government vehicles and machinery were vandalised. Hundreds were injured and hospitalised, while a similar number were arrested and locked up.
That episode exposed a contradiction of today’s times. Man has created machines capable of generating knowledge, language, images and decisions at speeds previously unimaginable. But even as AI (Artificial Intelligence) rewrites economies and societies globally, the very people tasked with ensuring that it stays safe are abandoning ship. All in the name of that safety being shunned for profits, cash and money.
This is no fringe rebellion. It is unfolding at the heart of the world’s most powerful AI laboratories. In 2026 alone, hundreds of top researchers tasked with AI safety and alignment – those who ensure AI systems remain controllable, ethical and beneficial – have resigned from companies like OpenAI and Anthropic. Their exit has not been quiet. For they have publicly warned that commercial ambition is overtaking sensibility, and that a mad race for dominance is eroding safeguards meant to protect society itself.
The irony is impossible to ignore. The architects of ‘Safe AI’ are suddenly turning whistle-blowers.
Silicon Rush
Dead-centre of this storm lies a cold business reality. AI is no longer just a technological frontier; it is the world’s largest capital magnet. Venture capital firms, sovereign wealth funds and global corporations are pouring massive sums into AI companies, rewarding speed, disruption and market capture, above all else.
Anthropic, recognised as one of the more safety-conscious AI firms, recently secured funding that pushed its valuation into staggering dollar territory. Almost on the same dates as the funding came in, it saw senior-level exits from researchers working on safety and alignment. One departing researcher, Mrinank Sharma, warned in a public note: “The world is in peril.” The message carried enormous weight. When those responsible for studying AI’s risks themselves express despair over its safety, society has to pause.
OpenAI is also witnessing turbulence around its safety structure. Its ‘Superalignment’ initiative – studying long-term AI risks – was dissolved earlier. Recently, its Mission Alignment team was broken up. Former OpenAI scientist Jan Leike said: “Safety culture and processes have taken a backseat to shiny products.”
India’s Exposure
India cannot afford to treat this as a distant Silicon Valley drama. It is among the world’s fastest-growing digital societies, with AI rapidly integrating into governance, education, banking, healthcare, policing, media and e-commerce. It is this extraordinary scale that makes India vulnerable to the unintended backlash of unrestrained AI deployment.
The dangers are already visible. AI-generated misinformation has become sophisticated and harder to detect. Deepfake videos targeting politicians, celebrities and even ordinary citizens have proliferated across social media platforms. Cybersecurity threats are intensifying as AI enables automated fraud, phishing and identity theft. The World Economic Forum has identified AI as the single-largest disruptor in cybersecurity, with others warning that vulnerabilities in AI are among the fastest-growing threats.
For India, the implications are profound. In a country where crores are entering the digital economy for the first time, even minor breaches of trust can trigger large-scale societal consequences. Then comes the question of privacy. AI systems thrive on data – oceans of it. Every conversation, facial image, financial record and behavioural pattern becomes raw material for machine learning models. If regulation is weak, this creates conditions ripe for surveillance, profiling and abuse.
To its credit, the Indian Government has recognised the gravity of the challenge. The Digital Personal Data Protection framework, advisories on deepfakes and interventions on intermediary accountability signal an emerging awareness that AI cannot remain a free-for-all technological frontier. Yet, with innovation accelerating at a frightening speed, policy is struggling to keep pace.
Human Cost
Beyond cybersecurity and misinformation lies another crisis: the erosion of humans. AI is replacing or reshaping white-collar work that was considered immune to automation. Coders, designers, analysts, writers and customer-support employees are now competing against systems capable of performing tasks in seconds. Even the creators of these technologies admit that future disruptions will be deadly.
The danger is not just job loss. It is social destabilisation to. When societies adopt new technologies faster than they develop ethical, educational and regulatory cushions, inequality wins. Wealth concentrates. Trust fractures. Democracies become vulnerable to manipulation. Human beings lose visibility into the systems governing their lives. This is why the resignation of safety researchers matters so profoundly.
These researchers are not anti-tech activists. They are insiders. Individuals who helped build the very systems now reshaping civilisation. Their departures suggest not opposition to innovation, but fear that humanity is surrendering oversight for speed.
The Last Guardrail
Today’s world resembles the early years of the Industrial Revolution, only vastly faster and infinitely more consequential. India, with its demographic heft, tech ambitions and expanding digital infrastructure, is at a critical crossroads. It cannot afford either reckless techno-utopianism or paralysing fear. Instead, it must pursue a third path: innovation plus accountability. That means AI audits, transparency norms, data-protection tools, reporting of AI-related incidents and guardrails for AI systems.
ALSO READ: The Melodi moment in Rome carried heavy geopolitical messaging
It also means ensuring that safety researchers are empowered, not marginalised. It means recognising that public trust, once lost, is far harder to rebuild than any machine. At the end of the day, AI may become man’s greatest invention. But if the guardians of AI safety walk away in fear, we must confront a terrifying possibility: perhaps the race is not about building intelligent machines, but about whether our wisdom can keep pace with ambition. That could be a far more difficult challenge.
The writer is a veteran journalist and communications specialist.