Amodei, known as one of the most critical voices for the uncontrolled development of AI, writes that the world is “much closer to real danger” in 2026 than it was in 2023, when the debate over the existential risks of AI intensified politically.
Humanity is entering a phase of artificial intelligence development that will “test who we are as a species,” the CEO of the company Anthropic has warned, urging the world to “wake up” to the approaching dangers. Dario Amodei, co-founder and CEO of the startup behind the chatbot Claude, has expressed his concerns in a nearly 19-word essay titled “The Adolescence of Technology.” In it, he writes that the arrival of extremely powerful AI systems could be imminent.
“I believe we are entering a rite of passage, a turbulent and inevitable process that will test who we are as a species,” Amodei writes. According to him, humanity is about to take on “an almost unimaginable power,” while it remains unclear whether society, politics, and technology are mature enough to handle it. Amodei emphasizes that his essay aims to “shock” public opinion and spur concrete action on the safety of artificial intelligence. The warning comes at the same time that the British government announced that Anthropic will help create chatbots for public services, including assistance for job seekers and career advice. Anthropic, founded in 2021 by former OpenAI employees, has also recently published an 80-page “constitution” for Claude, where it explains how it aims to make artificial intelligence “broadly safe and ethical.”
Amodei, known as one of the most critical voices about the uncontrolled development of AI, writes that the world is “much closer to real danger” in 2026 than it was in 2023, when the debate over the existential risks of AI intensified politically.
He also mentions the controversy over sexual deepfakes created by Elon Musk’s Grok chatbot, which spread massively on the social network X during the holiday season, including warnings that child sexual abuse material was being created. According to Amodei, the negligence of some companies to these problems raises serious doubts about their ability to manage even greater risks in the future. According to him, very powerful AI systems, which can act autonomously and even build new systems on their own, could appear within one to two years. He defines “strong AI” as an intelligence smarter than a Nobel Prize winner in fields such as biology, mathematics, engineering and writing, capable of collaborating with humans, controlling robots and even designing them.
While acknowledging that this timeline could be longer, Amodei stresses that the rapid progress of recent years should be taken seriously. “If exponential development continues, it could take no more than a few years for AI to be better than humans in almost every field,” he writes.
Amodei has previously warned that artificial intelligence could halve entry-level jobs in the administrative sector and push unemployment to 20% within five years. In the essay, he warns that the economic benefits of AI, such as increased productivity and job losses, could be so great that no one would dare to set limits. “This is the trap: AI is so powerful and so seductive that it is very difficult for human civilization to put any brakes on it,” he writes. However, Amodei is optimistic that a positive outcome is possible. According to him, if we act carefully and decisively, the risks can be overcome and beyond them there may be a much better world. “But we must understand that this is a serious challenge for civilization,” he concludes.