Anthropic warns that it’s very close to RSI capability, aka self-building AI

Anthropic, an AI lab, is close to achieving recursive self-improvement (RSI), a point at which AI grows so good at training itself that it spawns a dramatic acceleration of AI capabilities. RSI is seen as a double-edged sword, promising unimaginable productivity and scientific discovery, but also posing a safety risk if it accelerates progress too quickly. Anthropic uses AI in its R&D and has already used AI to develop new models with unanticipated leaps in capability, and may soon meet its own definition of RSI.