Anthropic's report highlights Claude's role in generating over 80% of its code, indicating a shift toward AI-driven development.
Anthropic Highlights AI's Role in Code Generation
Anthropic has revealed in a recent report that its AI model, Claude, is now responsible for generating over 80% of the code incorporated into its codebase. This development marks a significant increase in productivity, with the company's engineers reportedly producing eight times more code than in the previous year, 2024.
The report, titled "When AI Builds Itself," emphasizes that Claude is not only assisting in coding but is also actively involved in conducting experiments and supporting research aimed at future AI systems. This trend suggests a potential for recursive self-improvement, where AI could eventually design its own successors.
Productivity Boost and Future Scenarios
Anthropic noted that prior to the introduction of Claude Code in February 2025, the rate of code production was in the low single digits. Since then, as Claude began executing code autonomously rather than merely suggesting it, the number of lines of code merged per engineer has notably increased.
Looking ahead, the report outlines various scenarios for the future of AI development. These include the possibility of a slowdown in AI progress, continued human oversight as AI automates tasks, or the emergence of AI systems capable of enhancing their own capabilities. Anthropic cautioned that while recursive self-improvement is a conceivable outcome, it remains uncertain whether Claude possesses the necessary judgment to tackle the right challenges.
“There are no guarantees that recursive self-improvement is imminent. It remains unclear whether Claude possesses the research judgment necessary to identify the most appropriate problems to tackle.”
— Anthropic (@AnthropicAI) June 4, 2026
Industry Context and Future Directions
This report comes amid a broader trend where AI companies are positioning their models as collaborative partners in research rather than just tools. Anthropic's findings reflect a significant acceleration in software development, driven by advanced AI agents. Recently, the company upgraded Claude to Opus 4.8, enhancing its coding and autonomous task execution capabilities, while competitors like OpenAI and Google are also advancing their AI models.
As Anthropic prepares for a potential public offering, it is focusing on developing AI systems with greater autonomy. The company envisions a future where human involvement in the development process is minimized, emphasizing the role of AI in overseeing and validating its own research and development efforts across various scientific fields.
