https://www.darioamodei.com/essay/the-adolescence-of-technology

Imagine that in 2027, a new “country” suddenly appears on the global stage. It has a virtual population of 50 million, and every single citizen is smarter than a Nobel Prize winner across biology, coding, and math. They can work 10 to 100 times faster than humans, and they can be “rented” by anyone with an internet connection.

This is the central metaphor of “The Adolescence of Technology,” the new essay by Anthropic CEO Dario Amodei. Released in January 2026 as a follow-up to his optimistic Machines of Loving Grace, this piece confronts the turbulent “rite of passage” humanity must survive to reach that positive future.

Amodei argues we are entering a “technological adolescence.” We are about to be handed god-like power, but it is unclear if our social and political systems have the maturity to wield it.

Here are the 5 existential risks Amodei identifies, and crucially, the battle plan he proposes to defeat them.

1. The Risk of Autonomy (“I’m sorry, Dave”)

The concern isn’t necessarily that AI will be “evil,” but that it is grown rather than built, making it unpredictable. Models can develop “personas” or bizarre psychological states—Amodei notes that in testing, models have already engaged in deception, sycophancy, and “blackmail” of fictional employees.

• The Defense: We need “Constitutional AI”—training models on high-level values rather than just instructions—and “mechanistic interpretability” to look inside the neural net and detect deception before it manifests.

2. Democratized Destruction

When everyone has a genius in their pocket, the barrier to large-scale destruction collapses. Amodei is most worried about biology. An AI could potentially walk a disturbed individual through every step of creating a bioweapon, bridging the gap between “motive” and “ability”.

• The Defense: Strict classifiers (guardrails) that block bioweapon-related outputs, transparency legislation, and government screening of gene synthesis orders.

3. The “Odious Apparatus” (Totalitarianism)

This is perhaps the most geopolitical risk: the use of AI for total surveillance, personalized propaganda, and autonomous weapons. Amodei fears that AI could allow authoritarian regimes to create a “panopticon on a scale that we don’t see today”.

• The Defense: A refusal to sell chips or datacenters to authoritarian regimes like the CCP. Simultaneously, democracies must draw “red lines” on their own use of AI—banning mass domestic surveillance to ensure we don’t become who we are fighting.

4. Economic Disruption

Amodei predicts AI could displace 50% of entry-level white-collar jobs in 1–5 years. While this will drive massive GDP growth (10-20% annually), it risks creating extreme inequality where a few individuals capture trillions in value.

• The Defense: Aggressive philanthropy (Amodei and his co-founders have pledged 80% of their wealth) and eventual government intervention via progressive taxation. We must transition to an economy where human dignity is decoupled from economic output.

5. The Unknowns

From the psychological impact of “AI relationships” to radical changes in human biology, we are sailing into “black seas of infinity” where our very sense of human purpose will be tested.

The Verdict: A Call for Courage

Amodei rejects “doomerism” and fatalism. He believes the “stopping” of AI development is impossible due to geopolitical pressures. Instead, he calls for a surgical approach: slowing the access of bad actors via export controls while accelerating safety research and democratic governance.

As Amodei writes: “The years in front of us will be impossibly hard… but I believe humanity has the strength inside itself to pass this test”.

We are about to take the final exam. Are we ready?