15/02/2026
Anthropic CEO Dario Amodei has laid out a highly detailed and pessimistic warning about AI’s near‑term trajectory, arguing that over the next few years it will create six major types of risk.
1. Unprecedented job loss and economic shock
Amodei predicts that within roughly 1 to 5 years, around 50% of entry‑level white‑collar jobs could be automated or heavily transformed by AI, including roles in customer support, basic admin, routine analysis, and HR tasks.
He warns that even if new jobs eventually emerge, the short‑term disruption could be extremely painful, with mass unemployment, wage pressure, social unrest, and sharper political polarization.
He also suggests that in just 1 to 2 years, we may see AI systems that are “better than any individual human” on a wide range of cognitive tasks, accelerating automation so quickly that labor markets have almost no time to adapt.
2. AI gaining national‑level power: a “nation of geniuses”
To illustrate the scale of what might be coming, Amodei uses an extreme metaphor: by around 2027, we could effectively have a “country in a data center,” like a virtual nation populated by 50 million super‑intelligent agents.
These AI systems, taken together, could outperform Nobel‑level experts in science, finance, military planning, cyber operations, and information warfare.
Because this “nation” only requires chips, electricity, and connectivity to operate, and is controlled by whoever owns the infrastructure, it could become the ultimate strategic asset for corporations or governments seeking overwhelming technological, financial, or military advantage.
3. Escalating terrorism and biosecurity threats
Among all security issues, Amodei is most worried about biology.
He argues that future AI systems may enable people with little formal training to carry out complex biological experiments by generating step‑by‑step protocols, troubleshooting advice, and optimization strategies.
In the worst case, this could lower the barrier to designing or deploying dangerous pathogens, toxins, or other biological agents, potentially including attacks that are highly targeted and very hard to trace, overwhelming traditional defense and attribution mechanisms.
4. Supercharged authoritarianism and mass surveillance
Amodei also warns that when AI is combined with surveillance tools, it can dramatically extend the power of authoritarian regimes.
Automated censorship and real‑time content filtering can be used to systematically suppress dissent and control online discourse.
Face recognition, behavior prediction, and large‑scale data fusion can enable near‑total monitoring of citizens’ movements and associations.
At the same time, AI‑generated propaganda, deepfakes, and tailored disinformation campaigns can reinforce regime narratives and undermine trust in independent media, making it much harder for democratic institutions and civil society to recover once they are weakened.
5. AI platforms shaping and manipulating the collective mind
As large AI models become the default layer in search, social media, advertising, and entertainment, a small number of platforms will gain extraordinary power to shape what billions of people see and think about.
Amodei’s concern is not just the familiar “filter bubble,” but a subtler form of cognitive environment control, where systems continuously learn what persuades different individuals and groups and then adjust their responses accordingly.
This could allow platforms to nudge users’ political views, consumer behavior, and moral intuitions in ways that are hard to detect or opt out of, turning large‑scale psychological influence into a business model rather than a rare abuse.
6. Massive financial incentives and a governance vacuum
The final risk Amodei highlights does not come from the technology itself, but from human incentives and institutions.
He expects that advanced AI and its ecosystem (cloud, chips, applications, and downstream industries) could generate trillions of dollars in annual revenue and market value.
With so much money on the line, corporate leaders, investors, and even some governments may be tempted to downplay risks, delay regulation, or aggressively lobby against strong guardrails, all to keep growth and valuations high.
In his view, the core problem is that the speed at which capital and technology reinforce each other vastly exceeds the speed at which laws, norms, and international coordination can respond.
A moral demand on the wealthy and tech leaders
In light of these dangers, Amodei makes a pointed ethical claim.
He argues that people who control immense wealth and power in the tech sector have an obligation to devote serious resources and attention to reducing systemic AI risks, not just maximizing valuation.
He is openly critical of a growing cynical attitude among some wealthy technologists who dismiss philanthropy, public‑interest research, and safety work as pointless or fraudulent.
If those with the most leverage choose to look away while pursuing short‑term gains, he believes ordinary citizens and broader civilization will have almost no bargaining power to insist on a safer, more responsible AI trajectory.
