Artificial intelligence is rapidly developing from a technological competition into one of the most consequential political and economic debates of the decade, as some of the people building the world’s most advanced AI systems warn that the race itself may be moving too quickly.
Anthropic CEO Dario Amodei has called for a global slowdown in the development of advanced artificial intelligence, with the company indicating it is prepared to reduce the pace of development even if competitors do not follow. OpenAI CEO Sam Altman has advocated “pacing” rather than completely pausing AI development, while Google DeepMind CEO Demis Hassabis and Elon Musk have also expressed support for slowing aspects of the race.
One of the central concerns is that increasingly capable AI systems are becoming better at assisting humans in the development of the next generation of artificial intelligence. That could accelerate technological progress further, potentially reducing the amount of time governments, companies and researchers have to understand and manage emerging risks. Anthropic and OpenAI are also providing outside evaluators with greater access to their systems as part of efforts to strengthen independent safety monitoring.
But calls for greater regulation are facing skepticism in Washington. White House AI adviser David Sacks has suggested that established AI companies could use safety concerns to support regulations that ultimately strengthen their market position and make it more difficult for smaller competitors to challenge them.
That leaves policymakers confronting two competing risks: whether increasingly advanced AI could become dangerous enough to justify slowing development, and whether restricting American companies could allow China or other competitors to gain a strategic advantage. President Donald Trump has emphasized the geopolitical stakes of the competition, declaring during recent remarks in Ireland that “whoever wins AI wins.”
The issue is also moving deeper into electoral politics. Democratic strategists recently gathered in Washington for an extensive discussion about artificial intelligence and its potential role in the 2028 presidential campaign. Proposals under consideration included restrictions on data-centre development, greater liability for AI companies and possible limits on advanced AI systems.
For some Democrats, the political argument is shifting beyond whether artificial intelligence itself is safe. They are increasingly focusing on who receives the economic benefits of the technology and who absorbs its costs. That could turn AI into a populist economic issue involving employment, wages, education, energy prices and the concentration of wealth and technological power.
Artificial intelligence is also creating political alliances that would normally appear unlikely. Progressive Senator Bernie Sanders and MAGA commentator Steve Bannon are participating in a gathering advocating greater human control over advanced AI. Republican Senator Marsha Blackburn, progressive Congressman Greg Casar, former Anthropic researcher Jacob Coxon, faith organizations and other public figures are also involved.
The emerging coalition suggests the politics surrounding artificial intelligence may not fit comfortably within traditional Republican-versus-Democratic divisions. Figures on both the political left and right have expressed concerns about concentrated technological power, potentially creating a different dividing line between those who want rapid AI development and those demanding stronger limits and human oversight.
At the same time, the physical infrastructure required to support artificial intelligence is turning the debate into an issue that could directly affect households. Texas has halted approvals for new data centres amid concerns over their enormous electricity requirements, a notable development for a state known for its business-friendly policies and significant technology investment.
Advanced AI requires vast computing resources, which in turn require increasingly large data centres. Those facilities create demand for additional electricity generation, transmission infrastructure, substations and cooling systems. Congress is also examining whether ordinary electricity customers should be expected to absorb some of the infrastructure costs associated with the rapid expansion of data centres.
That could ultimately transform artificial intelligence from an abstract debate over algorithms and technological risk into a kitchen-table political issue. Many voters may never need to understand precisely how a large language model is trained, but they will understand changes in their electricity bills, employment prospects and local infrastructure.
The emerging AI debate therefore extends far beyond which company or country develops the most powerful model. Governments are increasingly being forced to decide how quickly the technology should advance, who should regulate it, who should pay for the infrastructure it requires and who ultimately receives its economic benefits.
The political battle may no longer divide neatly along conventional party lines. Instead, one of the defining divisions of the AI era could increasingly be between those who believe winning the technological race requires moving faster and those who believe the consequences are becoming significant enough to slow it down.
