“AI Is About to Test Who We Are as a Species”

In late January 2026, something unusual happened in Silicon Valley. One of the people actively building the most powerful AI systems on Earth stepped forward and said, plainly: we may be in danger. Dario Amodei — co-founder and CEO of Anthropic — did not issue this warning as an outsider, activist, or academic skeptic. He spoke as an insider. As the leader of a company behind Claude Opus 4.5, and whose internal AI agents now reportedly generate a large share of Anthropic’s production code, Amodei operates at the frontier of modern artificial intelligence. And his message was stark. In a 38-page essay titled “The Adolescence of Technology,” Amodei argues that humanity is approaching a period in which it may be handed “almost unimaginable power” — and that our institutions, economies, and moral frameworks are deeply unprepared for what could follow. This is not science fiction. It is not a speculative thought experiment. It is a warning from someone with direct visibility into how quickly these systems are advancing. As someone who has spent more than a decade helping banks, insurers, and asset managers deploy machine learning in real-world settings, I’ve learned to be skeptical of grand technological claims. But Amodei’s argument is neither vague nor sensationalist. It is precise. It is grounded in observed trends. And it is urgent. If the people building frontier AI systems are openly worried about civilization-scale failure modes, the rest of us should pause — and pay attention. Humanity’s Technological Adolescence Has Begun Amodei frames the current moment with a metaphor borrowed from Carl Sagan’s novel Contact . In the story, humanity asks a more advanced civilization how it survived its “ technological adolescence ” — the dangerous period when a species gains the ability to destroy itself before developing the wisdom to manage that power. Amodei suggests that humanity may now be entering such a phase . Artificial intelligence is not just another general-purpose technology. It is a force multiplier for intelligence itself. And intelligence, unlike labor or capital, can scale recursively: systems that are good at problem-solving can be applied to improving t…