What execs and politicians are saying about slowing down AI development

AI development leaders and politicians are actively debating and proposing different approaches to regulating AI, moving the conversation from theoretical risks to concrete policy discussions.

Anthropic CEO Dario Amodei initiated a significant discussion on AI safety by publishing an essay advocating for a slowdown in AI development, proposing three key steps: embedded third-party evaluators for safety verification, inter-company coordination on standards, and global governmental collaboration. Amodei stated Anthropic is unilaterally committing to the first step, emphasizing that "pacing" does not mean halting progress but rather ensuring adequate time for model alignment and safeguarding. OpenAI CEO Sam Altman expressed agreement with Amodei, endorsing the idea of independent evaluators and welcoming a federal regulatory framework for AI safety. He also echoed Amodei's sentiment that pacing means slowing progress, not stopping it, and highlighted risks of losing control to AI or excessive power concentration. Google DeepMind chair Demis Hassabis and Elon Musk also voiced support for Amodei's direction. In contrast, former President Donald Trump criticized the calls for regulation, asserting his administration's control over AI and warning against a conspiracy targeting AI development. US Vice President JD Vance expressed skepticism about tech companies seeking government regulation, framing it as a potential "Trojan horse" amid an AI race with China. House Speaker Mike Johnson acknowledged the need for balance between national security and AI safety, suggesting a meeting with industry leaders, while Senator Bernie Sanders endorsed the idea of pausing AI for humanity's sake.

The statements from leading AI executives like Amodei and Altman, supported by industry figures such as Hassabis and Musk, indicate a unified push from major AI labs towards establishing external oversight and industry-wide safety standards. This coordinated approach suggests a proactive effort to shape regulation before it is imposed, potentially influencing future governance frameworks and embedding specific industry-preferred safety mechanisms. Anthropic's unilateral commitment to third-party evaluation signals a potential competitive differentiator or a first step toward a de facto industry standard.

Conversely, political reactions reveal a divided landscape, with some, like former President Trump and VP Vance, expressing distrust of AI companies seeking regulation and emphasizing nationalistic competition, while others, like Speaker Johnson and Senator Sanders, engage with the safety concerns, albeit with different priorities. This divergence highlights the challenge of achieving global consensus on AI governance, especially when national security and economic competitiveness are perceived as being at stake, creating potential friction points for policy implementation.