Anthropic CEO Dario Amodei is calling for a slowdown in AI development, warning that advances in AI could soon outpace current safety measures.
OpenAI CEO Sam Altman and xAI CEO Elon Musk have backed the idea.
The White House has pushed in the other direction, arguing that the U.S. cannot afford to give up its lead over China and saying, “whoever wins AI wins.”
The debate is already reaching markets. AI stocks fell Monday as investors considered what a slower pace could mean for spending across the industry, while China rejected the proposal and pushed back on the warnings coming from U.S. AI leaders.
Why It Matters: Companies have committed large budgets to AI based partly on expectations that models will keep getting more capable. A slower development cycle could affect plans for chips, cloud capacity and data center infrastructure. However, the extra time could give researchers more room to understand new risks and strengthen safeguards before more capable models reach production.
- Warnings From Inside Anthropic: Amodei’s proposal comes as current and former Anthropic researchers raise questions about whether safety work can keep pace with more capable models. Former researcher Jacob Coxon cited concerns about advanced AI when he resigned, while Anthropic researcher Evan Hubinger has put the chance of AI causing human extinction within the next decade at more than 10%. Anthropic’s recent misuse report also documented Claude being used for cyber operations, surveillance, fraud and weapons development. Amodei has separately warned that within 6 to 12 months, more capable AI could potentially direct large groups of agents across the internet.
- More Time Between Advances: The proposal would give researchers an additional 1 to 2 years to improve protections while AI continues to advance. It also calls for independent evaluators to receive employee-level access to review AI systems, an approach Altman said OpenAI would adopt. Altman has called for companies to coordinate on safety without waiting for governments to act, and Musk has publicly backed the proposal.
- A Sell-Off Across AI: Calls for slower AI development sent shares lower across companies tied to chips, infrastructure and data centers. SoftBank fell about 10% in Japan, while SK Hynix dropped more than 6% and Samsung Electronics lost more than 4%. The sell-off extended into Europe, where ASML, Nokia and Infineon declined, followed by losses for Micron, Intel and Nvidia in U.S. trading. Investor expectations for these companies remain closely tied to how quickly AI development moves and the infrastructure spending that comes with it.
- The U.S.-China Race: Trump has rejected slowing U.S. development because of competition with China, saying the country should maintain its current lead. He said guardrails could still be put in place, while arguing that some of the risks being raised will not materialize. Beijing has also rejected calls for a slowdown, describing the warnings as “fear mongering.” Amodei has raised the same competitive concern, warning that holding back U.S. developers without similar action elsewhere could give Chinese companies time to catch up.
- Inference Keeps the Pressure On: Slowing the development of new models may have a limited effect on the computing power companies already need. Training accounts for only part of that demand, since running AI models in production also requires substantial capacity. Analysts say inference capacity remains in short supply, which could sustain spending on chips and AI infrastructure even if the next generation of frontier models takes longer to arrive.
Go Deeper -> AI stocks slide after major CEOs unite to urge slowdown – CNBC
US tech stocks fall after calls for AI development to slow down – BBC
New warnings about the risks of AI to humanity revive a long-running debate – AP News
China says AI CEOs’ call for a slowdown is ‘fear mongering’ – CNBC
