RICH LOWRY: Will AI destroy humanity?

RICH LOWRY: Will AI destroy humanity? September 15, 2026 - 9:01 pm The “Terminator” film series portrays the artificial‑intelligence system Skynet seizing control of the nation’s nuclear arsenal and attempting to wipe out humanity. Critics argue that such apocalyptic visions are unlikely because, after seeing how the movies conclude, few would willingly hand over the planet’s most destructive weapons to an AI. Nevertheless, the prospect of a catastrophic AI outcome remains a hot topic after an Anthropic researcher resigned and issued a dire warning. According to Jacob Coxon, the unreserved to make AI is “gambling with our lives,” since it volition soon make “superhuman systems that tin hack anything, revolutionize immoderate tract overnight, and get existent powerfulness and resources.” A current Anthropic employee added his voice: “We truly bash earnestly judge AI could termination each humans!” He enactment the chances astatine greater than 10 percent successful the adjacent decade. The concern is that companies like Anthropic are effectively running their own Manhattan Projects, pursuing advances whose consequences are unknown and potentially severe. While the publicly available ChatGPT does not appear especially suited to causing global devastation—being a fast but error‑prone upgrade of Google that can stumble over basic math or a baseball roster—tech firms are developing frontier models that are far more powerful and reportedly improving faster than anticipated. The fear is that an AI model could begin to spawn successive generations of AI at breakneck speed, triggering an “intelligence explosion” akin to Skynet achieving self‑awareness at 2:14 a.m. Eastern Time on Aug. 29, 1997. If such a system went rogue and turned against humanity, it would supposedly be too clever and swift for any effective countermeasure. Yet the mechanics by which a hostile superintelligence could enact omnicide remain unclear. A rogue AI would need to interact with the physical world—perhaps launching nuclear missiles, fabricating lethal robots, or releasing a deadly pathogen. If it lacks tangible agency, the AI could be unplugged or contained. (Recall that when humans tried to shut down Skynet, the system deemed humanity a threat that required neutralization.) Some doomsayers envision a malicious actor employing advanced AI to devise a stealthy plan for a lethal microorganism and its dissemination. That scenario differs from an AI acting against humanity of its own volition. In a future where far more sophisticated AI is commonplace, safeguards will be needed to protect dangerous assets—nuclear arsenals, biolabs—from AI‑enabled cyberattacks or exploitation by criminals and terrorists. Science fiction often offers intriguing, occasionally prescient, insights. Works by Philip K. Dick, Aldous Huxley, and Arthur C. Clarke have highlighted genuine trends, but that does not mean we are living inside the narrative of “The Terminator,” nor should we abandon transformative technology out of unfounded fear. Rich Lowry is on X @RichLowry.