More than a quarter of Anthropic’s AI research and development led by Claude
Anthropic reports that its Claude chatbot now drives 26 percent of its AI research and development work. This is a significant increase from near zero at the start of the year. Claude also collaborates with staffers on 90 percent of their work. The company released these figures in a report addressing concerns about AI moving beyond human control.
Anthropic's 26 percent figure for Claude leading R&D is a stark indicator. AI systems are now actively building themselves. This recursive self-improvement (RSI) is not just theoretical. It is a present reality for frontier AI labs. The company also uses over 30,000 agents for research and engineering work.
The implications for Asia's AI sector are direct. Companies like South Korea's Naver or Japan's Preferred Networks must now consider this accelerated development cycle. Relying solely on human-led R&D will fall behind. The race to integrate AI into model development itself intensifies. This will push Asian firms to invest more in internal AI tooling.
The core tension remains between speed and safety. Anthropic allocates only 6 to 12 percent of computing resources to safety monitoring. This is a small fraction compared to development. Asian regulators face pressure to enforce safety standards without stifling innovation. This balance is critical for the region's AI future.
Related reading
6 storiesGemini 3.8 Live: Google Still Trails OpenAI, Anthropic - Memeburn
We previously covered how Google's Gemini still trails OpenAI and Anthropic in AI model development.
Google Shipped Gemini 3.8 Live — and It Still Feels Like Third Place
Our earlier analysis noted Google's Gemini 3.8 Live release still felt like third place behind competitors.
China makes progress in 3-nm chips without advanced lithography tools
In India, the new iPhone can arrive faster than a pizza
Grok releases Voice Transcribe 2.0 with improved accuracy
