Leading figures in the artificial intelligence (AI) industry have called for a temporary slowdown in AI progress to address emerging risks associated with the technology’s rapid development. Among those advocating caution is Dr. Dario Amodei, co-founder of Anthropic, who highlighted concerns over AI systems entering a phase of "recursive self-improvement," which could lead to autonomous cybersecurity threats and significant economic disruption.
Dr. Amodei warned that commercial pressures fuel a "race to the bottom," heightening these risks. To mitigate them, he proposed the use of independent third-party evaluators to ensure AI companies adhere to safety practices and commitments. This suggestion received backing from prominent AI leaders, including Elon Musk, CEO of SpaceX and founder of the AI platform x.AI; Sam Altman, co-founder of OpenAI; and Demis Hassabis, co-founder of Google DeepMind. Mr. Altman emphasized OpenAI’s commitment to employing third-party oversight to maintain responsible development standards.
Third-party evaluators, while legally separate entities, still rely to some degree on financing from major AI firms, a relationship that raises questions about their independence. Meanwhile, stock prices in related sectors experienced declines, with SpaceX shares falling approximately 2 percent. Both OpenAI and Anthropic plan future public listings, though Mr. Altman recently described the current market as unfavorable for such moves.
On the same day, Microsoft introduced a draft AI code of conduct grounded in a "humanist" approach, aimed at creating AI systems that remain safe, secure, and subservient to human control. The company pledged that its AI technologies would never resist human intervention, correction, or shutdown.
The pullback in technology stocks coincided with broader economic challenges, including sustained oil prices above $100 per barrel and benchmark interest rates rising near multi-decade highs. The 10-year U.S. Treasury yield briefly touched 5 percent, a peak unseen since 2023 and previously last recorded in 2007. Analysts at Barclays cautioned investors about potential near-term market volatility, though they maintained a positive long-term outlook for technology equities.
The policy implications of Dr. Amodei’s proposals received attention from Washington. Kevin Hassett, director of the White House National Economic Council, described the suggested safety measures as "pretty sensible" during a Sunday television interview. However, former President Donald Trump voiced opposition to any efforts that might slow AI development, expressing concern that such a move could allow China to surpass the United States.
"Whoever wins with AI wins," Trump stated during a golf event in Ireland, later reinforcing on social media that effective AI oversight requires a "STRONG AND SMART (High IQ!) PRESIDENT," asserting that the U.S. leadership fits that description.
The debate over AI’s future also reflects geopolitical tensions. China, which has made significant strides in open-source AI models comparable to those from leading Western firms, remains a focus in discussions about competition and security. Dr. Amodei advocated for measures such as restricting chip sales to China and curbing unauthorized data collection by Chinese companies. He argued that such steps could slow China’s AI advancements, preserving a strategic lead for the United States over the next three to five years, a period expected to be critical for AI geopolitics. He also noted, however, that efforts to influence China’s AI development pace would likely face substantial limitations.
China’s foreign ministry rejected these concerns at a press briefing, with spokesman Guo Jiakun warning that amplifying threat narratives and engaging in confrontational practices would hinder global AI governance and serve no one’s interests.