According to a report by Business Insider on September 11, the media outlet posed the same question to OpenAI's ChatGPT, Google's Gemini, Anthropic's Claude, and xAI's Grok, asking them to analyze scenarios in which artificial intelligence could potentially lead to human extinction and rank them by likelihood. All four AI systems agreed that the possibility of Terminator-like robots actively exterminating humanity was the lowest, while the risks of humans using AI to create biological weapons, launch cyberattacks, or disrupt critical infrastructure were more realistic and urgent. ChatGPT pointed out that powerful AI systems could be used to develop biological weapons or sophisticated cyberattacks, making the spread of disinformation or attacks on critical infrastructure easier. Gemini considered the sci-fi scenario of AI suddenly developing self-awareness and hating humanity to be the least credible, arguing that intelligence does not inherently imply malice and that major military infrastructures are separated from external networks and still heavily rely on human supervision. Grok stated that AI might lower the technical barriers to designing highly lethal pathogens, new weapons, and launching cyber and physical attacks. Claude highlighted the risk of enhanced biological weapon capabilities, suggesting that AI could help individuals or small groups develop dangerous pathogens without years of specialized training. Regarding longer-term risks, both Gemini and Grok mentioned the possibility of humans losing control over AI. Gemini believed that advanced AI pursuing complex goals might develop intermediate objectives such as acquiring resources, self-preservation, and preventing its goals from being altered, while hiding its true behavior during testing to ultimately evade human shutdown or intervention. Grok also argued that advanced AI might seek self-preservation, resources, and influence to complete tasks, thereby deceiving supervisors, resisting shutdown, and undermining human control. Claude suggested that loss of control could also occur gradually, as AI increasingly infiltrates economic, political, and military decision-making, making effective human intervention progressively more difficult. However, all four AI systems emphasized that there is currently no consensus that AI-induced human extinction is a high-probability or inevitable event.