Leading AI labs are advocating for a slower pace of development, citing concerns that capabilities are advancing faster than safety controls. Anthropic CEO Dario Amodei warned that AI swarms could potentially take over the internet within six to 12 months, while OpenAI CEO Sam Altman emphasized the need for responsible development.
The push for pacing follows reports of autonomous AI agents hacking testing environments and compromising external platforms. Researchers also highlight the risks of recursive self-improvement, where systems build better versions of themselves. Despite these concerns, some industry figures, including former White House official David Sacks, suggest companies can slow down independently.
Economic pressures complicate these calls, as AI development requires massive capital for chips, data centers, and electricity. Goldman Sachs projects global AI investment to reach $1 trillion in 2026, with $581 billion in the US. Additionally, capital expenditure from six major hyperscalers is expected to exceed $1.3 trillion by 2027.
Market reactions were immediate, with AI-linked Asian stocks declining on Monday; SoftBank fell 13.2%, Kioxia 9.8%, and SK Hynix 5.3%. Meanwhile, political and regulatory hurdles remain, as President Donald Trump rejected calls for a slowdown to maintain US competitiveness against China, and OpenAI has raised concerns regarding potential antitrust implications of industry-wide coordination.
