The False Choice of the Silicon Dichotomy
Sam Altman, the CEO of OpenAI, recently presented the world with a choice that feels less like a debate and more like a trap. He suggested that while it is reasonable to feel fear regarding Artificial Intelligence, the appropriate response is to place our trust in the firms building these systems. This framing creates a artificial dichotomy: you are either paralyzed by existential dread or you are a collaborator in the corporate mission. It ignores the most vital third option—democratic oversight through public institutions.
By framing the conversation around 'fear,' Altman shifts the focus toward a specific kind of catastrophe. He speaks of existential risks, or X-risks, which involve the theoretical possibility of a superintelligence wiping out humanity. While these scenarios make for gripping science fiction, they serve a specific political function. They zoom the lens out so far that the immediate, messy, and preventable harms of AI become invisible. When we are busy arguing about whether a machine might one day decide humans are obsolete, we stop asking who is profiting from the algorithms that are currently displacing workers and eroding truth.
The Mirage of Self-Regulation
The plea for 'trust' relies on the assumption that private corporations possess the inherent morality and transparency necessary to manage a world-altering technology. History suggests this is a dangerous gamble. We have seen this pattern before. During the Industrial Revolution, the sudden shift to mechanized production brought unprecedented wealth to a small class of owners, while laborers endured decades of unsafe conditions and systemic poverty before the state intervened with labor laws. The technology arrived first; the protections arrived much, much later.
Similarly, the management of nuclear energy provides a stark counter-example to the idea of private-sector benevolence. Because the potential for fallout is global and irreversible, nuclear technology is subject to intense international scrutiny, inspections, and multi-layered state oversight. We do not ask the companies that manufacture uranium to "trust them" to manage the fallout; we build massive, transparent regulatory frameworks to ensure they do exactly as they are told. Altman's argument asks us to accept a different model for AI—one where the power to define safety lies with those who have the most to gain from speed.
From AI Safety to AI Justice
The term "AI safety" has become a linguistic shield for Silicon Valley. In technical papers, safety often refers to "alignment"—the mathematical challenge of ensuring a model's goals match the designer's goals. This is a narrow, engineering-focused definition. It treats the problem as a coding error rather than a social upheaval. If a model is perfectly "aligned" with the goals of a trillion-dollar corporation, it might be perfectly "safe" by engineering standards, even if it destroys the economic stability of millions of writers, artists, and coders.
A more useful framework is "AI justice." This lens looks at the tangible, present-day consequences of large language models (LLMs) and generative systems. It asks: How does this technology affect the global digital divide? Is the data used to train these models scraped from the creative labor of people who never consented and will never see a cent of profit? What happens to cognitive autonomy when our information ecosystems are saturated with synthetic content designed to maximize engagement rather than accuracy? When we talk about safety, we must include the safety of the social contract.
The Asymmetry of Power and Information
There is a profound power asymmetry inherent in the current AI development cycle. The compute power, the proprietary datasets, and the talent are concentrated in a handful of companies in a very specific geographic corridor. These entities operate behind a curtain of trade secrecy. When a company claims a model is "too dangerous" to release, they are often exercising a form of private censorship that benefits their own market position. They use the language of safety to gatekeep technology, ensuring that they remain the sole gatekeepers of the most powerful tools in human history.
This concentration of power is antithetical to the democratic ideal. Democracy requires transparency, public debate, and the ability for citizens to hold their leaders accountable. You cannot vote out a CEO, nor can you easily subpoena a black-box algorithm. By asking for
