Anthropic's research arm publishes an update on its work toward recursive self-improvement in AI systems. The article discusses progress toward systems that can autonomously improve their own capabilities, a key research frontier in AI safety and capability development.