Anthropic says Claude is building its own successor

Anthropic says it is using Claude to help build future Anthropic models, according to the original report. OpenAI has previously disclosed that it uses its own language models in chip design. Taken together, these are early signs that leading AI companies are putting their models to work on parts of the process that creates the next generation of models.

That does not mean AI is independently designing, training, or releasing its own replacement. People, compute infrastructure, training data, testing, and company decisions still sit around that work. But it does make the idea of recursive improvement more concrete for buyers. If AI tools can speed up research, coding, testing, documentation, or system design inside frontier labs, capability gains may arrive faster than many planning cycles assume.

For a business owner, the practical question is not whether to panic about AI building AI. It is whether your technology plan can handle a faster rate of change. Avoid locking critical workflows into one model, one vendor, or one set of assumptions about what AI can do. Track where your teams use AI, what data it touches, and which processes would be affected if model quality or pricing shifts quickly.

This is the kind of thing we build into an AI command centre, so leaders can see AI activity, risk, and business value in one place rather than reacting after the fact.