Anthropic has, for the first time, made public its three key metrics regarding artificial intelligence (AI) research and development (R&D), confirming that RSI (AI self-evolution) has indeed taken place within its labs. To begin with, Claude has assumed a leading role in 26% of the core R&D tasks, with deep human-machine collaboration being achieved in over 90% of the R&D process. Furthermore, the company is currently operating around 30,000 R&D agents on a daily basis internally. A two-tiered monitoring system—comprising real-time online interception and offline monitoring with human final review—along with independent identities and a transparent broadcasting framework, ensures the security of R&D activities. Additionally, Anthropic has shared its computational power allocation data, indicating that only approximately 6% of the core R&D computational power is dedicated to safety research. The company clarified that safety research is more reliant on human intellect and that this figure represents a conservative estimate. Anthropic's initiative is designed to bridge the information divide between leading-edge laboratories and the general public. It urges the entire industry, particularly rivals such as OpenAI, to release comparable data to boost transparency within the sector and empower society to make informed decisions on how to address the current state of AI self-evolution.