Anthropic says self-building AI may soon reach a major milestone, raising urgent questions about how quickly the technology could transform jobs. In an amended filing with the Securities and Exchange Commission on November 2, 2026, the company said future models could meet its definition of dramatically accelerated research progress powered by AI. Anthropic also made clear that no current Claude model has reached that threshold.
The technology is known as recursive self-improvement, or RSI. Essentially, it describes a future in which artificial intelligence becomes capable of improving the systems that power it, potentially creating increasingly advanced versions with less human assistance. Anthropic already uses Claude to help write code, conduct research, run experiments, and develop newer models. However, the company says meaningful human supervision remains necessary.
For millions of workers, the concern is what happens if self-building AI becomes powerful enough to perform more complicated tasks faster than businesses can train employees to keep up. Companies would no longer have to wait as long for human developers to create each technological improvement. More capable systems could help produce their successors, potentially accelerating workplace automation along the way.
And workers may feel the consequences long before fully autonomous AI becomes reality.
Consider entry-level positions. Businesses traditionally hire assistants, junior developers, customer service representatives, and administrative employees to handle tasks while gaining experience. AI can already perform portions of that work. If companies increasingly assign those responsibilities to software, they may reduce new hiring rather than immediately lay off existing employees.
Research suggests younger workers could be especially vulnerable. A June 2026 report from Stanford University’s Digital Economy Lab found that employment among workers ages 22 to 25 in highly AI-exposed occupations was declining at an annual rate of 3.8% in its sample, while employment in the least exposed occupations was growing by 2%. The findings do not establish AI as the sole cause, but they reveal a concerning divide in early-career opportunities.
The potential reach of self-building AI extends well beyond technology companies. According to the International Labour Organization, approximately one in four workers worldwide holds a job with some exposure to generative AI. Clerical occupations face particularly high exposure, while increasingly sophisticated systems could also reshape specialized professional work.
Made to order and shipped to your door.
Customer service agents, graphic designers, marketing employees, writers, financial analysts, and software developers could see more responsibilities automated. That does not mean every position will disappear. However, a company that once needed several employees to complete a project might eventually accomplish similar work with a smaller team supported by AI.
This creates another potential problem: fewer opportunities to develop valuable skills.
A junior programmer becomes a senior developer by solving problems, making mistakes, and working alongside experienced professionals. If self-building AI takes over more beginner assignments, companies could eventually struggle to develop the human expertise needed to supervise increasingly complex systems. Workers could also find themselves competing for positions requiring experience they were never given the opportunity to earn.
The effects could reach freelancers and small businesses, too. A business owner might use AI to create advertisements, manage customer inquiries, analyze sales, and develop marketing materials. Those savings could help smaller companies compete, but they could also reduce demand for independent contractors who previously earned income providing those services.
There is also the question of wages. If employers can accomplish more with fewer people, workers in certain fields could lose bargaining power. Businesses might offer lower pay for tasks that AI can perform cheaply, while employees face increasing pressure to master new tools simply to remain competitive.
The pace of self-building AI could make that adjustment even harder. Traditional technological changes often unfold over years, giving schools, businesses, and workers some time to respond. AI systems capable of accelerating their own development could shorten those adjustment periods significantly.
Anthropic has acknowledged another risk: its models could advance faster than the company can develop and validate adequate safety measures. If increasingly autonomous systems make decisions affecting hiring, finances, or customer services, errors and biased outputs could affect real people. Human accountability becomes especially important when automated decisions are difficult to understand or challenge.
Still, widespread unemployment is not a guaranteed outcome. AI could also create jobs, increase productivity, support scientific discoveries, and allow employees to focus on responsibilities requiring human judgment. The World Economic Forum’s 2025 Future of Jobs Report projected 170 million new positions and 92 million displaced positions globally by 2030, reflecting several economic and technological forces, not AI alone.
The central issue is whether new opportunities will emerge quickly enough, in the right places, for the workers whose jobs change or disappear.
Employers could reduce some of the damage by retraining existing staff, maintaining entry-level career pathways, and giving employees meaningful oversight of automated systems. Governments and educational institutions may also need to invest in affordable training, worker protections, and stronger accountability standards.
As self-building AI becomes more capable, the benefits of increased productivity could flow largely to the companies that own the technology unless workers share in those gains.
For employees trying to build stable careers, the concern is no longer limited to whether AI can do their jobs today. It is whether the next generation of AI will become capable of doing those jobs before they have a realistic chance to prepare.
