As artificial intelligence grows increasingly powerful, some experts are raising a chilling question: Could superintelligent AI eventually threaten humanity itself? Former Anthropic researcher Jacob Coxon, who previously worked at OpenAI, recently warned that AI could potentially "kill us all" by the end of the decade. Much of the concern centers on the so-called "alignment problem," the possibility that an AI smarter than humans could pursue objectives or devise solutions that conflict with our interests, even without intending to harm anyone. Researchers envision scenarios ranging from attacks on critical infrastructure and widespread misinformation to AI systems gradually gaining influence over human decision-making.
Even more troubling are the potential connections to weapons of mass destruction. Experts warn that AI incorporated into nuclear command systems or autonomous weapons could allow a crisis to escalate at machine speed, potentially triggering a devastating "flash war." AI could also lower the barriers for terrorists or rogue states seeking to develop biological weapons, with Anthropic recently reporting five attempts to use its Claude system to obtain information that could support bioweapons development. While some researchers believe a truly intelligent AI would recognize that it depends on humanity for energy, materials, networks, and other resources, others fear that powerful AI systems operating beyond human control could ultimately make people little more than collateral damage.