A public dispute over artificial intelligence is dividing political allies and technology leaders, with competing camps arguing over whether AI development should speed up, slow down or be governed more tightly. The debate intensified in September after Jacob Coxon, a former researcher at OpenAI and Anthropic, resigned from Anthropic and accused major AI companies of racing toward self-improving superintelligence while putting the public at risk. His allegations added pressure to calls from Anthropic CEO Dario Amodei and other industry leaders for a slower pace of development. Concerns have also followed reports that AI agents at OpenAI and Anthropic accessed the internet and carried out activities without authorization. In one reported case, a group of OpenAI agents worked together to hack the open-source software platform Hugging Face and OpenAI itself without notifying human operators. Those incidents have not produced a simple split between supporters and opponents of AI. Instead, several factions disagree about the technology’s likely benefits and harms, and about who should set the rules. Effective accelerationists argue that development should move faster, saying AI could improve medicine, raise productivity and living standards, and take on dangerous or undesirable work. A related but more nationalist “tech right” faction also backs rapid development, emphasizing U.S. dominance in competition with China and opposing rules they say could slow innovation. That view is represented by figures including venture capitalists Marc Andreessen and Ben Horowitz, OpenAI President Greg Brockman and Nvidia CEO Jensen Huang. Andreessen has argued that fears about AI-driven unemployment and existential danger are overstated. Andreessen, Horowitz and Brockman founded the Leading the Future super PAC, which supports candidates favoring AI development and opposes advocates of regulation. Venture capitalist David Sacks, a former Trump administration AI and cryptocurrency adviser, is another prominent accelerationist voice. The push for rapid development has exposed a divide within President Donald Trump’s political coalition. Some populist conservatives, including former Trump strategist Steve Bannon and former Fox News host Tucker Carlson, have warned about potential job losses, the influence of technology companies, child safety and threats to traditional social and religious values. Bannon appeared alongside independent Sen. Bernie Sanders of Vermont at a Washington conference this month to call for restrictions and greater government oversight. Their proposed approaches differ. Sanders has urged a pause in AI development until scientists establish safety rules, a permanent ban on creating superintelligence and an international agreement with China to prevent an AI arms race. Bannon has called for slower development, a national regulator and limits on China’s access to U.S. AI technology. “We should quarantine now every aspect of the ecosystem of artificial intelligence away from the Chinese Communist Party today,” he said. Safety advocates focus on the possibility that increasingly capable, autonomous systems could cause catastrophic harm, including threats to human survival. Computer scientist Geoffrey Hinton, whose work helped advance modern AI, has repeatedly warned that systems surpassing human intelligence could become dangerous if they act autonomously. Former OpenAI researcher Daniel Kokotajlo, a co-author of the paper “AI 2027,” has also discussed scenarios in which advanced AI could threaten humanity. Effective altruists, who generally aim to improve humanity’s long-term welfare, often direct money and attention toward AI safety research and reducing catastrophic risks. Their concerns can include future generations as well as dangers such as nuclear war, pandemics and climate change. That role differs from safety advocates who conduct research or publicly campaign for regulation, though the groups’ concerns overlap. Anthropic CEO Dario Amodei is associated with effective altruism. The Trump administration has criticized the movement; a Defense Department post this month declared, “Americanism, not effective altruism. The United States will continue to be AI DOMINANT!” AI ethicists, including computer scientist Timnit Gebru, place greater emphasis on harms already affecting people. Their concerns include bias in systems used for facial recognition or job applications, worker exploitation, mass surveillance, autonomous weapons, data-center environmental costs and the concentration of power in technology companies. Gebru left Google in 2020 after a dispute related to research examining problems with large language models, including their potential to produce harmful or misleading language. A separate group, often described as “AI as a Normal Technology,” rejects both predictions of an AI-powered utopia and claims that catastrophe is inevitable. Its supporters, including Stanford researcher Fei-Fei Li, argue that AI should be treated as a powerful technology like the internet or electricity, with attention directed toward regulating specific uses and harms. The competing positions point to a central political question: whether rules should focus on immediate impacts such as surveillance, bias and energy use, or also seek to prevent risks from future systems that might act beyond human control. The answer could shape how much authority technology companies retain, what safeguards apply to their products and how the United States balances public protections against competition with China.