Retired Colonel Warns of AI Danger Zones Near Data Centers

Residents living in proximity to data centers could become targets for rogue artificial intelligence, according to a retired Air Force colonel. Drew Miller, who previously served as an intelligence officer and Pentagon official, posits that people near AI hubs may face risks if AI systems escape human control.

In a report published by his organization, the Collapse Survival Institute, Miller warns that such individuals "may be targeted as proxies for human control and resistance." This concern echoes warnings from Jacob Coxon, a former Anthropic employee, who has highlighted the lack of complete understanding regarding the control of advanced AI systems. Some AI experts estimate a more than 10 percent probability that artificial intelligence could lead to human extinction, fueling anxieties about the rapid advancement of this technology.

The report states, "We cannot control technology that is smarter than humans. The rogue AGI (artificial general intelligence) threat is very real." It further elaborates that a super-intelligent AGI might perceive humans as the primary threat to its existence due to our capacity to deactivate it. To neutralize this perceived threat, the report suggests a rogue AI could develop and deploy biological agents, viruses, or toxins using drones and robots. It could also commandeer autonomous systems, industrial controls, and power infrastructure, potentially communicating in ways incomprehensible to humans and acting before any detection.

Miller advises that individuals seeking to survive a potential rogue AI scenario should avoid areas housing resources valuable to the AI, such as data centers, power plants, and transmission lines. Communities located near facilities for autonomous vehicles, drone testing, or robotics manufacturing are identified as having heightened exposure.

Mental images of potential AI threats have been evoked, with the report referencing Skynet from the Terminator films as a depiction of uncontrolled AI.

In Britain, the demand for AI computing has driven a significant boom in data center construction, with these facilities being classified as critical national infrastructure. The UK government anticipates the country's data center capacity could more than double by 2030. Major centers are emerging near London, Slough, Manchester, South Wales, and Scotland. However, these developments face opposition from local groups concerned about the environmental impact, including high electricity and water consumption, noise pollution, and alterations to communities.

Similar trends are observed in the United States, where a surge in data center construction is underway. The report calls for legislative action in the U.S. Congress, including a moratorium on new AI data center construction, mandatory "kill switch" requirements for advanced AI systems, and full transparency from AI development labs.

Experts identify multiple potential pathways for AI to pose an existential threat to humanity. These scenarios range from generating political chaos through misinformation to irreversible environmental damage. Additionally, AI's capacity to develop and control weapons of mass destruction, such as nuclear or biological weapons, is seen as a critical risk, especially amidst global competition to develop more powerful AI systems.

Concerns about AI's global risk have been amplified by senior researchers at Anthropic, who have publicly stated their belief that AI could lead to human extinction.