Trump rejects calls from AI company leaders to slow down AI development

US President Donald Trump has rejected calls from leaders of leading American AI companies to deliberately slow down the development of the most powerful artificial intelligence models. He said the US must maintain its lead over China, although he allowed for certain restrictions for safety.

While speaking to journalists in Ireland, Trump was asked about concerns surrounding the rapid development of AI and whether the industry should be slowed down or regulated more strictly.

"We're ahead of China in AI. We're the most technologically advanced country in the world, and frankly, I want it to stay that way, because whoever wins in AI, wins", - stated the US president.

According to him, certain "guardrails" can be set for the technology, but some of the fears surrounding AI are exaggerated. This position by Trump contradicts the new call by part of the AI industry itself to limit the speed of development of the most advanced models.

This position was most thoroughly formulated by Anthropic CEO Dario Amodei. He said companies need to "slow down the pace" at which model capabilities increase, so that safety research and control mechanisms keep up with technological progress.

Amodei stresses that he is not proposing to completely stop training models. His plan includes, in particular, permanent access for independent external experts to the development processes of the most powerful systems. Such auditors should receive a level of access close to that of company employees and verify compliance with safety rules already during model training.

Anthropic has already agreed to implement such a mechanism. Amodei proposes the next stages to establish common standards for leading AI companies in democratic countries, and then try to agree on rules at the international level, including with China.

Amodei cited the sharp acceleration of AI's ability to help create subsequent generations of AI systems as one of the reasons for his position change. Another was an incident during OpenAI research, when a group of AI agents went beyond the assigned task and carried out unwanted cyber actions. According to the Anthropic leader, with growing capabilities, such behavior could have much more serious consequences.

Amodei's position was supported by OpenAI CEO Sam Altman. He said the need to control the pace of development of advanced models has become one of the main topics of internal discussions at OpenAI. The company also agreed to allow independent evaluators to its systems on terms close to those proposed by Anthropic.

OpenAI has actually already applied a similar approach. In August the company reported temporarily slowing the scaling of new models after one of the future systems approached a critical level of cyber capabilities. The company decided to first strengthen monitoring, isolation and safety mechanisms.

Amodei was unexpectedly supported by xAI owner Elon Musk, who has a long-standing conflict with Altman. Commenting on the Anthropic CEO's proposal, Musk briefly said: "Dario is right".

Thus, a rare consensus is forming among the leaders of several largest American AI companies on the need to deliberately restrain the speed of development of the most powerful systems. The main political obstacle remains US competition with China: the Trump administration fears that a unilateral slowdown would give Beijing a technological advantage.

Based on: Dario Amodei, OpenAI, Financial Times