AI building AI? Anthropic warns future is closer than you think

AI may soon help create its own successors, but what does it mean for AI Developers? Traditionally, human researchers and engineers were deeply involved in building AI models. But Anthropic says it has started to assign more of the development tasks to AI systems themselves. The company highlighted that “Taken far enough, and given enough compute,” AI systems will autonomously be able to create their own successors.

Must read: Anthropic expands Claude Mythos access to India; Adds 15 other countries and 150 organisations

AI will soon develop AI

According to an Anthropic blog post, AI is advancing at a rapid speed, and soon it will eventually help develop the next generation of AI on its own. The company talked about the “Recursive self-improvement” process, where an AI system helps build AI designs, trains and improves on its own.  The AI model will also be able to test and identify improvements and create an even better AI system. However, it said, “We are not there yet, and recursive self-improvement is not inevitable.”

“AI that can build itself would be a major development in the history of technology—one that could bring enormous good for the world in science, healthcare, and beyond,” the blog says. However, it also highlights the risks of “humans losing control over AI systems,”  and “ it may also compound alignment issues and ultimately lead to loss of control.”

Must read: Trump’s new AI order could change how OpenAI, Google, and Anthropic launch models

In addition, Anthropic says the world should have the ability to temporarily slow or pause the development of advanced AI systems if the technology progresses faster than safety measures, regulations, and oversight frameworks can adapt.

Currently, the technology is being leveraged to help people write, code, research, and automate tasks. Now, it’s also helping researchers build better AI models. However, the capabilities have expanded drastically into AI models helping employees write entire code, finding and fixing bugs, conducting research, and testing new ideas. Now, as AI models get more powerful, they will soon be able to assist in building even more advanced models.

Anthropic also acknowledged that there are many unanswered questions about AI, including fully recursive self-improvement, where AI could enhance its own capabilities. It plans to have further discussions with researchers, policymakers, and the public to discuss these possibilities and concerns.