Anthropic warns that it’s very close to RSI capability, aka self-building AI
Anthropic's filing warns it may soon meet its internal definition of recursive self-improvement, where AI dramatically accelerates its own development.Summary
In its revised S-1A filing, Anthropic disclosed that it may be on the cusp of recursive self-improvement (RSI), the point where AI becomes so good at training itself that capabilities accelerate dramatically. The company added new risk statements saying it has already used AI to develop models with capability leaps it didn't recognize at the time, and that its internal estimates indicate it may imminently meet its Responsible Scaling Policy definition of "dramatically accelerated progress attributable to AI." However, Anthropic said no current model has crossed this threshold yet and promised to keep humans in the R&D process for the foreseeable future, considering updates to its policy to maintain human control. While skeptics note that an S-1 serves as both disclosure and marketing, Claude's capabilities in coding, math, and cybersecurity have grown rapidly this year, suggesting the warning deserves serious attention.
(Source：TechCrunch)