AI labs push toward autonomous self-improvement, with Anthropic and OpenAI leading the charge

Leading AI labs say artificial intelligence models are nearing "recursive self-improvement," where AI can autonomously enhance itself, raising both promises of scientific advances and fears about AI evading human control. Anthropic reported its Claude model now leads 26% of its research and development, completing tasks end-to-end under human supervision. OpenAI announced an automated "research intern" and aims for a fully automated AI "researcher" by March 2028. Experts note some self-improvement has occurred for years, but fully autonomous leaps remain untested. Elon Musk predicts xAI's Grok could achieve full automation by 2027, while Microsoft pursues a more limited "humanist superintelligence" approach.