A new controversy is brewing around the development of artificial intelligence: Anthropic, OpenAI, and Elon Musk almost simultaneously began talking about the need to slow progress on advanced models. The reason was cited as the risk of recursive self-improvement, but critics also argued it was an attempt to protect the positions of key players.
Anthropic CEO Dario Amodei said that artificial intelligence has begun to help create the next generation of systems, which is why the development speed is so fast that people will lose control of the process. The company intends to independently limit the speed of development and allow third-party experts to check the safety of the model and learning process itself. The idea quickly gained support from Sam Altman and Elon Musk.
This synchronicity raises a lot of suspicions. Opponents of the initiative argue that strict regulation will create high barriers mainly for smaller companies and open-model developers, while strengthening the positions of current market leaders. This seems particularly controversial against the backdrop of the success of Chinese projects such as DeepSeek, Moonshot and Z.ai, which are increasingly competing with US systems.
Another interesting thing: there are now rumors that Google DeepMind may be making significant progress in recursive self-improvement of artificial intelligence. If this is true, systems that can help create one’s own successors could significantly speed up competition. In this case, the call to slow down is not only for safety reasons, but also to gain time.