Senator **Bernie Sanders** recently ignited a critical discussion in the *New York Times*, posing a profound question: "Will the future of humanity be determined by a handful of billionaires who have promoted and developed AI, with virtually no democratic input, who stand to become even richer and more powerful than they are today?" This question resonates deeply within the cybersecurity and privacy community, as the concentration of power, wealth, and control among tech oligarchs presents an urgent risk to global democracy. ### The Allure of a Sovereign AI Fund Sanders advocates for a seemingly radical, yet increasingly popular, solution: establishing a U.S. sovereign wealth fund. This fund would acquire a 50% stake in major AI companies such as **Anthropic**, **OpenAI**, and **xAI**. The senator's argument is twofold: 1. **Democratic Control:** This would grant the government significant voting shares and board representation, enabling it to "block decisions that hurt our citizens and to push for policies that help them." 2. **Wealth Redistribution:** It would ensure that the "trillions of dollars potentially generated by AI are used to improve the lives of all of us," returning a substantial portion of AI's economic rewards to the public. These goals, ensuring public influence and disrupting the accumulating wealth of AI giants, are widely supported. Companies like **OpenAI** and **Anthropic** are racing towards trillion-dollar valuations, a phenomenon that represents a massive transfer of wealth from smaller businesses and individual users to the owners of these tech behemoths. ### Unintended Consequences of Public Ownership Despite the appealing objectives, the concept of direct public ownership of AI companies through a sovereign fund introduces significant risks. Critics argue that such a model could inadvertently entangle corporate profit motives with public interest, creating perverse incentives for the government. This could lead to: * **Regulatory Leniency:** The government might be incentivized to clear regulations that benefit its corporate holdings. * **Exploitation Tolerance:** Pressure to maximize returns could lead to overlooking worker and user exploitation. * **Suppressed Competition:** Government-backed companies might face less competitive pressure. * **Uncritical AI Adoption:** The fund's managers might push for AI adoption regardless of ethical considerations or appropriate use cases. Historical examples support these concerns. The **Norwegian sovereign wealth fund**, despite its substantial stakes in oil companies, has not demonstrably steered those corporations toward pro-environmental policies. Instead, the government's financial dependence on these companies has, at times, inhibited climate action. Similarly, U.S. public employee pension funds, while holding corporate stakes, often prioritize fiduciary duty over public interest directives. ### Alternative Paths to Public Interest AI Rather than direct ownership, experts propose separating the goals of wealth redistribution and public interest influence. Two main alternatives are highlighted: #### 1. Taxation for Wealth Redistribution The most straightforward method to share private rewards with society is through taxation. Senator **Elizabeth Warren** has proposed an excise tax on data centers' energy use, while others suggest an AI token tax. These mechanisms ensure that AI companies contribute financially to the societal infrastructure and public services that enable their success, without the government becoming a direct stakeholder in their profitability. #### 2. The AI Public Option for Influence To reshape AI development in the public interest, an **AI Public Option** is proposed. This concept involves governments—federal or state—establishing publicly developed and operated AI models run by public institutions under democratic control. The aim is not to eliminate private AI but to provide a competitive baseline that private offerings must meet or exceed in terms of transparency, ethics, and public benefit. Switzerland offers a compelling example with **Apertus**, a large language model built by Swiss public servants and university researchers. Utilizing public supercomputing infrastructure and appropriately licensed data, Apertus, while not matching the raw performance of models from **OpenAI** or **Anthropic**, excels in transparency, sustainability, and compliance with regulations like copyright. This demonstrates how public initiatives can exert competitive pressure on corporate actors to behave more responsibly. It is crucial to distinguish this from "sovereign AI," a term often used by big tech companies like **Nvidia** to market domestic AI infrastructure investments to governments without guaranteeing public control. ### The Political Calculus Sanders' pursuit of the sovereign wealth fund strategy, despite its potential pitfalls, might stem from perceived alignment with diverse political factions, including the **Trump administration** and some AI billionaires. However, this alignment raises a critical question: why are these billionaires open to such extraordinary intervention? The likely answer is that they anticipate favorable government policies and protections for their investments, outweighing any shares ceded. Ultimately, energy taxation offers a direct way to make AI companies accountable for the societal impact of their technologies. A public AI option, on the other hand, provides a non-monetary mechanism for governments to shape AI development, complementing direct regulation and offering a greater chance of influencing corporate behavior toward the public interest. As the debate continues, political leaders are urged to consider these alternative, potentially more effective, strategies.