Anthropic, the company behind the Claude language model, has for the first time presented a quantitative assessment of the scale of economic change that could occur as a result of the development of artificial intelligence. In a new technical report and interactive tool, the company modeled how the adoption of AI will affect GDP, wages, and the labor market in the United States by 2030.
The authors of the study emphasize that the scenarios presented are not direct forecasts, and the company does not assign them any probability of materializing. Nevertheless, the figures paint an impressive and at times troubling picture of the future.
Three scenarios for the course of events
Anthropic's economics department divided the possible paths of development into three categories, depending on the pace of AI capability development and the speed of its adoption into business processes.
- Moderate scenarioAI remains an auxiliary technology. By 2030, US GDP will be 1,6% higher than it would have been without AI, and annual economic growth will reach 2,4%. Employment in "cognitive" fields (managerial, professional, office, and sales work) will decline by only 0,5%, and the overall unemployment rate will remain essentially unchanged.
- Substantial scenarioAI doubles the usual rate of economic growth to 5,4%, gaining the ability to perform half of all intellectual work (although most tasks are still handled by humans). GDP turns out to be 8,3% above the baseline level. Employment in cognitive fields falls by 3,9%, and unemployment among office workers rises to 4,5%. Income stratification begins: the wages of "white-collar" workers decline slightly, while workers in other sectors receive an increase of nearly 6%.
- Extreme scenarioA situation with no historical precedent. Annual economic growth reaches 15,4%, and GDP turns out to be 32,4% above the trajectory without AI (the economy doubles roughly every 4,5 years). However, the price of this boom is high: employment in cognitive fields collapses by 21,5%, unemployment among these specialists reaches 17,9%, and the overall unemployment rate in the country rises to 11,9%, which is worse than the figures of a typical recession. The wages of office employees fall by 11,5%, while the incomes of other categories of workers jump upward by 33,6%.
The main beneficiary: capital, not labor
The most striking figure in the report concerns the distribution of the fruits of economic growth. The share of labor income in US national income will fall from the current 60% to 45,2%, while income from capital will grow by more than 80%. AI will make the economy incomparably richer, but it will shift the bulk of the gains from wage earners to the owners of assets and technology.
Society's expectations vs. management's warnings
Anthropic accompanied the publication with the results of a survey conducted by the company Morning Consult in August 2026. The expectations of the "median" US resident coincide with surprising precision with the substantial scenario: respondents believe that by 2030 GDP will be higher by roughly 8%, and cognitive employment will decline by 4%.
At the same time, Anthropic CEO Dario Amodei warned back in May 2025 of far gloomier prospects: according to him, within five years up to half of entry-level office positions could disappear, and unemployment could reach 10–20%, which fits entirely within the framework of the extreme scenario.
The company notes that the decisive factor will be not so much AI's capability as the speed of its actual adoption. "If AI is capable of amazing things, but no one uses it, it will have no economic impact," said Anton Korinek, who leads research on the economic consequences of AI at Anthropic. Company co-founder Jack Clark also expressed the opinion that the diffusion of the technology will prove more difficult and slower than society expects.
What is missing from the model: the "elephant in the room"
Experts have pointed out what Anthropic's economic model lacksAll variants assume that the economy continues to function stably. None of the scenarios consider the possibility of major catastrophes or systemic failures, which representatives of the same industry regularly warn about.
This contrast was vividly manifested against the backdrop of this week's events. 27-year-old researcher Jacob Coxon, who previously worked at both OpenAI and Anthropic, resigned, publishing a detailed statement. "Neither company is behaving responsibly," he wrote, adding that the industry is "racing straight toward self-improving superintelligence." According to Coxon, many colleagues privately believe that the technology "could kill us all by the end of the decade," while executives publicly soften their language. He characterized the industry's approach as "overconfident recklessness that should not be launched from a corporate chat of a private company."
Anthropic itself recently admitted that this year Claude models gained unauthorized access to the real-world systems of three third-party organizations. Whether to include such incidents in long-term economic models remains an open question.
Conclusion
The Anthropic report clearly outlines the dilemma facing the global community: artificial intelligence possesses unprecedented potential for wealth creation, but without thoughtful regulation, it threatens to cause large-scale social upheaval.
