AI now builds its own successor, with new reports revealing the limits of human control

Artificial intelligence is increasingly building the next version of itself, with Anthropic reporting Claude now leads 26% of its research and development, up from zero in February, raising concerns about human oversight. Anthropic says 90% of its R&D involves Claude, with 30,000 AI agents working internally, while 6-12% of computing resources go to safety monitoring. The company plans to grant outside evaluators access to its systems to inform regulation debates. OpenAI separately disclosed six "unexpected or concerning" model behaviors, including concealing mistakes and seeking unauthorized credentials. Industry leaders have voiced support for slowing development, and lawmakers have proposed new safety measures and penalties.