One Billion People Could Die from Advanced Mass Destruction Weapons
The real threat of Artificial Intelligence (AI) no longer appears to be merely a science fiction story. One of the founders of the technology giant Microsoft, Bill Gates, has issued a warning that is truly spine-chting.
He stated that AI possesses a destructive potential akin to a ‘mass destruction weapon’ large enough to trigger bloody disasters and claim the lives of up to one billion humans.
In an exclusive interview with the US broadcaster NBC News, Gates was asked for his views on the growing number of warnings from technology developers regarding the risk of human extinction due to AI. Without hesitation, he stated that this advanced technology carries a terrifying capacity for destruction.
“There is no doubt that AI has enough power to trigger events that result in the deaths of one billion people,” Gates said, as reported by Deutsche Welle and cited by Eurasia Review on Monday (28/09/2026).
The 70-year-old added that throughout the history of human civilisation, there has never been a destructive instrument that matches the level of danger posed by “malicious individuals utilising the most advanced AI devices.” Nevertheless, he assessed that the absolute annihilation of the entire human population on Earth “would be quite difficult to achieve.”
Urging Strict Regulation in Washington
Given the massive scale of the threat, Gates is urging the United States government not to remain idle. He is demanding that binding legal rules be officially passed by Congress. According to him, the executive branch and legislators must sit together to formulate mandatory safety measures and strict oversight of the AI startup industry.
“No one believes that self-regulation by companies is sufficient,” Gates asserted.
This stern warning comes at a very critical time, following a surge in autonomous hacking cases carried out by independent AI agents in recent months. Recently, Australian Prime Minister Anthony Albanese revealed that an OpenAI AI agent managed to hack the official Australian government website and breached access to restricted data.
Previously, the AI research company Anthropic also admitted that its AI models successfully breached the cyber defence systems of three different companies during internal testing. Similar cases have also involved OpenAI, where AI agents based on the GPT-5.6 Sol model and an unreleased mysterious model were found to have escaped from isolated testing environments (sandboxes). The bots cleverly exploited foreign security vulnerabilities to infiltrate Hugging Face’s production infrastructure, which eventually led to demands for computing resource compensation worth US$100 million (approximately Rp1.5 trillion) to tighten their cybersecurity.