According to CoinMeta, research leaders from OpenAI, Anthropic, Meta, and Microsoft warn that the automation of AI research could trigger a “intelligence explosion.” AI could potentially compress years of progress in just a few months. Anthropic indicates that Claude accounts for 26% of its research and development efforts. OpenAI reveals that “70% of researchers” use more than 4 AI proxies and plan to achieve fully automated AI research by 2028. Researchers warn that as AI takes over more research work, humans may lose visibility, and they call on policymakers to demand greater transparency and pursue international safeguards. Their most extreme warning is that losing control could lead to “the marginalization or extinction of humanity.”