When Sam Altman, the CEO of OpenAI, tells the world it is "right to be afraid" while simultaneously asking for blind trust, he is performing a delicate rhetorical balancing act. This narrative, echoed across Silicon Valley, sets up a false binary: on one side, there is the slow, cumbersome machinery of government regulation, and on the other, the rapid, innovative drive of private industry. By framing the conversation this way, tech leaders shift the focus away from democratic accountability and toward a choice between progress and stagnation. However, this debate is not actually about the speed of innovation versus the necessity of safety. It is about who holds the power to define what is safe, and who bears the cost when the technology fails.
The Fiction of Self-Regulation
The core of the industry's plea rests on the idea that companies can—and should—police themselves. Mark Zuckerberg has argued that profit motives act as a natural safety incentive; the logic being that if a product is dangerous or biased, users will leave, and the company will lose money. This ignores the reality of market dominance. When a few firms control the foundational models that power the digital world, they create ecosystems that are impossible to leave. In such a monopoly, profit motives often work in direct opposition to safety. A company incentivized by quarterly growth and market capture has every reason to deploy a model before its social harms are fully understood. Self-regulation in a competitive vacuum is a fantasy; in an oligopoly, it is a strategy to prevent oversight.
Deconstructing Alignment
Tech executives frequently use the term "alignment" to describe the process of making AI systems follow human values. It sounds benevolent, even universal. But the term masks a profound political question: whose values are we aligning to? When engineers in San Francisco or Seattle program the reward functions for Large Language Models (LLMs), they are encoding specific cultural norms, linguistic patterns, and ethical frameworks. These are often Western, neoliberal, and corporate-friendly values. "Alignment" risks becoming a tool for digital colonialism, where the moral consensus of a small group of engineers is exported globally, flattening the diverse ethical traditions of the world in favor of a sanitized, Silicon Valley standard of conduct.
Missing Perspectives and Digital Extraction
The current discourse on AI governance is remarkably provincial. The "doomsday" scenarios discussed by Altman and Jensen Huang—often involving rogue superintelligence—tend to occupy the center of the conversation, yet they rarely touch the lived realities of the Global South. For many, the danger of AI isn't an existential threat from a machine; it is the immediate reality of data extraction and labor exploitation. Massive amounts of low-wage human labor are used to label datasets and moderate violent content, often in countries with minimal labor protections. This is a new form of resource extraction where the raw material is human cognition and cultural data, processed by machines to generate massive wealth for a handful of corporations while leaving the source communities with little recourse.
The Specter of Existential Risk
Focusing on
