Anthropic Warns AI Could Soon Build Its Own Successors

AI Systems Move Toward Autonomous Development

Artificial intelligence development races forward at unprecedented speed. Anthropic now warns that AI systems could soon advance themselves without human involvement. The company released new research showing AI already helps build the next generation of models. This trend may lead to systems designing and improving themselves with minimal human guidance.

The AI startup shared its findings in a blog post from the Anthropic Institute. The company says the industry may reach “recursive self-improvement” sooner than expected. Recursive self-improvement describes a future where one AI model develops the next version of itself. Researchers still guide the process today. However, AI now handles a growing share of coding, debugging, and technical research inside the company.

Jack Clark, Anthropic’s co-founder and head of policy, addressed the timing. He wants lawmakers and institutions to understand what may come next. Clark emphasizes the importance of preparation. The company believes society needs time to adapt.

“We’ve always found that the best thing to do is to socialize the concept and basically give people a sense of what’s coming,” Clark said in a release.

Claude Generates Majority of Company Code

Anthropic revealed striking internal data about AI contributions. Claude-generated code now accounts for more than 80% of the code merged into Anthropic’s systems as of May 2026. Before Claude Code launched in early 2025, that figure sat in the low single digits. The transformation happened in roughly one year.

Engineering productivity has surged alongside those changes. Anthropic engineers now merge roughly eight times more code per day than they did in 2024. The company says lines of code merged per engineer per day stayed constant through Anthropic’s first four years. That period spanned 2021 through 2024. The metric began climbing upward in 2025 when Claude began running code rather than just suggesting it.

The shift represents more than incremental improvement. AI systems now complete increasingly complex assignments over longer periods without human intervention. Anthropic claims the length of tasks models can reliably handle has doubled roughly every four months. That acceleration suggests rapid capability growth.

Three Possible Futures for AI Development

Anthropic outlined several potential scenarios. AI progress could slow down. Humans might remain in charge while AI automates much of the work. Or AI systems could eventually begin improving their own successors. The company acknowledges uncertainty about which outcome proves most likely.

The report “When AI Builds Itself” argues that sufficient computing power could enable full autonomy. AI systems might gain the capacity to fully autonomously design and develop future versions. Anthropic emphasizes this moment has not yet arrived. Recursive self-improvement is not inevitable. But the company warns it could arrive sooner than most institutions expect.

Clark noted that AI progress appears to be accelerating instead of slowing. He said the shift could drive major gains in medicine, science, and other technical fields. Public benchmarks now track AI performance across software engineering and scientific research tasks. Those metrics show steady improvement.

Human Oversight Remains the Bottleneck

Anthropic suggests humans may now represent the biggest constraint. AI has become so effective at writing code and researching that human oversight limits development speed. The company says Claude already helps build future AI systems by writing code, running experiments, and assisting with research. That contribution grows larger every month.

The company acknowledges lines of code provide an imperfect productivity measure. It remains unclear whether Claude possesses true research judgment. The system may not yet choose the right problems to work on. But if current trends continue, AI systems designing and building their own successors becomes plausible.

Broader Implications for Society

The implications extend far beyond software engineering. If AI systems can reliably improve themselves, technological change could accelerate dramatically. Fields like medicine, physics, and materials science could see breakthroughs arrive at unprecedented speed. Society would need to adapt quickly to waves of innovation.

Regulatory frameworks designed for slower technological progress may prove inadequate. Governments typically take years to craft and implement new policies. AI development now moves in months or even weeks. The mismatch creates risks. Anthropic hopes to address those risks by raising awareness early.

The company’s transparency stands out in an industry often criticized for secrecy. Anthropic argues that socializing the concept helps society prepare. If recursive self-improvement arrives, institutions need time to adapt. Policies, safety protocols, and oversight mechanisms all require updates. That work takes time.

Uncertainty Remains About Future Pace

Anthropic’s warning echoes themes familiar from science fiction. Films and pop culture have explored AI self-improvement for decades. The difference now is that the possibility appears increasingly concrete. Internal metrics show AI contributing meaningfully to its own development.

Whether AI progress continues at its current pace remains uncertain. Technical barriers could emerge. Resource constraints might slow development. Safety concerns could require pauses. But evidence from Anthropic’s internal metrics suggests momentum remains strong. AI systems now contribute meaningfully to building future AI.

The company says recursive self-improvement is not guaranteed. Multiple factors could prevent it. But the trends point in that direction. Anthropic wants governments, researchers, and the public to understand the trajectory. Preparation matters more than prediction. Society needs frameworks ready for rapid change.

Lines of code represent just one metric. Other measures might tell different stories. But the eightfold increase in code output per engineer suggests real productivity gains. Claude now handles tasks that previously required human effort. That shift frees engineers to focus on higher-level problems.

The AI industry watches these developments closely. Anthropic positions itself as a leader in AI safety and transparency. The company’s public warnings aim to shape policy discussions. If AI systems soon design their own successors, society needs advance notice. That notice provides time for institutions to respond.