1 of 1
Story summary
- U.S. AI firms are advancing recursive self-improvement, prompting calls for a development slowdown.
- Anthropic reports its Claude model now performs 26% of the company's research and development work.
- OpenAI introduced a “research intern” program aiming to produce an AI researcher by March 2028, warning rapid RSI may be unsafe.
- Elon Musk said xAI’s Grok models will be built by preceding versions, targeting automation by end-2026, no later than 2027.