The narrative of a “golden age” for artificial intelligence (AI) is facing increasing skepticism. According to Barron’s, Stuart Russell, a computer science professor at the University of California, Berkeley and a leading authority in AI safety, recently told AFP at the AI Impact Summit in New Delhi that global tech giants are caught in an AI “arms race,” with potential consequences that could threaten all of human civilization.
“Allowing private companies to bet the fate of humanity in a game akin to ‘Russian roulette’ strikes me as a serious government failure,” Russell said.
At the same time, TechRadar points out that optimistic declarations around AI have persisted for years—industrial revolutions, efficiency leaps, frictionless productivity—but real-world evidence suggests that the risks are not minor issues solved by a single software update; they are structural challenges that grow with the technology.
Regulatory gaps and CEO dilemmas: Who will hit the brakes?
Russell believes that the heads of major AI companies are not unaware of the risks. On the contrary, they “understand the danger that superintelligent systems could surpass humans.” The problem is that no single company can unilaterally “disarm.”
“I believe every major AI CEO wants to end this race,” he said, “but they cannot do it alone, or they would be ousted by investors.”
Success
You are now signed up for our newsletter
Success
Check your email to complete sign up
He revealed that some executives have privately expressed similar concerns. Publicly, OpenAI CEO Sam Altman has also acknowledged the theoretical possibility that AI could lead to human extinction.
So far, international summits have mostly produced voluntary corporate pledges without binding enforcement. Russell emphasizes that real responsibility lies with governments taking collective action to establish hard regulatory frameworks.
Capital rush: data centers and $200 billion in investment
Beyond policy struggles, capital is moving at full speed.
Countries and tech companies are investing hundreds of billions of dollars in building high-energy-consuming data centers to train and operate generative AI models. India is attempting to leverage this trend. Indian IT Minister Ashwini Vaishnaw stated that over the next two years, India expects to attract more than $200 billion in AI investment, of which about $90 billion has already been secured.
But expansion also raises employment anxieties. India’s large customer service and tech support outsourcing industry has recently seen stock declines amid fears that AI assistant tools will replace many backend jobs. Russell bluntly stated, “We are creating ‘human imitators,’ and the most natural application, of course, is to replace humans.”
He also observed a backlash among younger people against AI “dehumanization.” “When systems take over answering questions, making decisions, and planning actions, you are effectively weakening human agency. Young people don’t want that future.”
Safety red flags: executive resignations, deepfakes, and real-world misjudgments
TechRadar has summarized several recent warning signals.
First is the wave of resignations in AI safety. An Anthropic executive in charge of AI safety research resigned publicly this month, warning that “the world is in danger.” Similar departures have occurred at labs like xAI. These exits are often accompanied by public statements questioning whether companies are sacrificing safety standards under competitive pressure.
Second is the proliferation of deepfake technology. Regulators have begun investigating platforms producing inappropriate images involving minors. As generative tools become easier to use, the technical barrier for creating fake images and videos has greatly lowered. The result threatens not just individuals but the foundations of public trust.
Additionally, the AI “hallucination” problem is moving from screens to the real world. Systems like autonomous vehicles, warehouse robots, and drones rely on visual recognition and real-time decision-making. Research shows that minor environmental disruptions or deliberately altered signs can cause system misjudgments—for example, interpreting a “STOP” sign as a “SPEED LIMIT” sign. What is a mistake in a lab could be a disaster on the road.
Commercialization boundaries: when chatbots meet advertising
Another controversial issue is business models. OpenAI has recently begun testing advertising products in ChatGPT, prompting concern among some researchers. A senior researcher resigned publicly, noting that when systems both understand users’ psychological profiles and carry commercial incentives, the line between assistance and manipulation becomes blurred.
The history of social media provides a cautionary example—ad-driven algorithm optimization often prioritizes engagement over well-being. If AI assistants incorporate revenue weighting into answer ranking and recommendation logic, the impact could be even deeper.
According to the AI Incident Database, between November 2025 and January 2026 alone, 108 new AI-related incidents were recorded, covering fraud, misleading advice, and system misuse. While some cases were small in scale, the growing number itself signals a trend.
