What changed
Anthropic disclosed in its revised S-1A filing that it is nearing the development of a model capable of recursive self-improvement (RSI). This capability would allow an AI system to significantly enhance its own abilities, potentially leading to rapid advancements in AI capabilities. The filing highlights that Anthropic has already used AI to develop new models, some of which showed leaps in capability that were not recognized at the time.
What this means in practice
The company acknowledges that using AI to build more capable AI could pose safety risks, particularly if it accelerates progress beyond current safety measures. Anthropic states that it is monitoring and mitigating these risks through its Responsible Scaling Policy. The filing notes that internal estimates suggest the company may soon meet its internal definition of RSI, which involves a dramatic acceleration in progress due to AI. However, none of its current models have crossed this threshold, and the company continues to assess the underlying data.
Limitations and availability
Anthropic emphasizes that it does not anticipate being able to safely use Claude to fully autonomously design and build successive generations of frontier models without meaningful human supervision. The company is considering updates to its Responsible Scaling Policy to ensure humans remain in control of frontier model development. The supplied announcement does not specify the exact timeline or technical details of when RSI capabilities might be achieved.
