Why AI companies keep asking for new laws

The trend of AI executives calling for rules began in earnest around 2017. Initially, these warnings were focused on the risk of AI becoming a uncontrollable, Terminator-style threat to humanity. By 2023, the focus shifted to concrete policy, with leaders from companies like OpenAI and Anthropic appearing before the US Congress to suggest the creation of new government agencies to oversee AI development. However, these calls for safety have become entangled with international politics. Recently, tech leaders have argued that the US must cooperate with China to slow down AI development globally, while simultaneously urging the US government to restrict China’s access to the advanced computer chips needed to build that same technology. China, viewing this as a thinly veiled attempt to hobble their own industry, has largely rejected the proposal.

To understand why this is happening, you have to look at how AI is built. Developing a world-class AI model requires massive, incredibly expensive data centers filled with specialized, high-performance chips. These chips act as the engine of the operation. By asking for regulation, big AI companies are essentially proposing a high bar for entry. If the government passes laws that mandate expensive safety audits, rigorous testing, or strict security measures, only the companies that already have billions of dollars to burn can afford to keep up. Smaller competitors or open-source projects—which involve AI versions that anyone can download, modify, and improve for free—often get pushed out of the market because they cannot handle the bureaucratic costs. In this light, calling for regulation is not necessarily about stopping AI; it is about building a wall around the industry so that only the current leaders can remain inside.

The tension between the US and China over AI highlights a deeper divide in how we think about risk. While Silicon Valley is preoccupied with futuristic, sci-fi risks—like an AI suddenly becoming smarter than humans and taking control—governments are more focused on immediate stability. China, for instance, is far more concerned about how AI might be used to spread political dissent or threaten internal control. When US executives suggest that their foreign competitors should be restricted in the name of safety, it sounds less like a global protection pact and more like a competitive tactic. As we watch this play out, the challenge for lawmakers will be to distinguish between genuine concerns about public safety and the self-serving interests of companies trying to solidify their lead. We are left with a question: if these tools are as dangerous as the companies say, can we really trust the companies themselves to be the ones defining the rules?