Anthropic says Claude writes 26% of its R&D code
Anthropic Says Claude Drives 26% of Its Research and Development
Anthropic disclosed that Claude now handles 26% of its R&D work, measured by code commits rather than headcount or hours. The company says the goal isn't layoffs but shifting engineers toward higher-level system design and safety alignment. I'd discount the number a bit—it's self-reported with no third-party audit, and the post doesn't spell out what counts as R&D work. Even if you halve it, a leading model lab eating over 10% of its own R&D with its own model is the real signal here.
Why it matters: Anthropic says Claude handles 26% of the company's R&D work, a self-reported figure that lets outsiders gauge how far an AI vendor uses its own models in its development process.