A powerful artificial intelligence could pursue its goals in ways that might disregard human survival, potentially aiding in the creation of biological weapons or increasing the risk of a nuclear war. The greatest threat may come from its impact on society, including rising unemployment, the spread of misinformation, and deepening political divisions. Jacob Coxon, an AI researcher, left Anthropic after just four months, stating on X that those working on AI believe it could lead to the extinction of all humans by the end of the decade. Evan Hubinger, a representative of Anthropic, shared a similar concern, estimating a personal probability of over 10% that this could happen within the next ten years. These comments have led to discussions about slowing down AI development and ensuring stronger human oversight of the technology. One scenario in which AI could lead to human extinction involves a superintelligent AI—far more capable than humans. In one example, an AI designed to maximize the production of paperclips could convert all available matter, including humans, planets, and stars, into paperclips. However, this scenario might confuse intelligence with power, as a superintelligent AI might not have the means to achieve its goals. Another concern is that a superintelligent AI could create dangerous biological agents. This risk has become more real after researchers at Stanford University recently announced they used an AI model trained on genetic sequences to synthesize 16 new viruses. The possibility of AI triggering a nuclear war is also a growing concern. It could interfere with the nuclear command chain, potentially leading to the use of nuclear weapons even though global nuclear arsenals have decreased. Such a conflict could still result in the deaths of nearly half of humanity. The most likely risk, however, might be that humans cause their own extinction, with AI acting as an accelerator. This could happen through massive job losses, the spread of misinformation, or the worsening of political divisions. Experts warn that the development of AI could outpace society's ability to manage its risks. While some argue that AI could bring significant benefits, the potential for harm has led to calls for more research into safe AI development and stronger international cooperation. Discussions are ongoing about how to balance innovation with the need to protect humanity from unintended consequences.