Carnegie Endowment | Scott Singer
U.S. President Donald Trump and Chinese President Xi Jinping are slated to discuss artificial intelligence, a critical agenda item given the escalating risks posed by frontier AI models like Claude Mythos and GPT-5.5, which have uncovered cybersecurity vulnerabilities and threaten financial systems. While past U.S.-China AI dialogues in 2024 failed due to misaligned objectives—U.S. focusing on technical risks and China on export controls—the strategic landscape has shifted. China has demonstrably increased its investment in AI safety, acknowledging extreme risks, making a focused dialogue on shared concerns now feasible. A critical first step involves prioritizing discussions on standardized testing and evaluation methods for extreme AI risks, including the use of "red teams," while carefully limiting the sharing of sensitive intellectual property or dual-use domain knowledge. This bilateral foundation is essential for establishing global AI safety protocols, despite the inherent challenges of U.S.-China diplomacy.
No comments:
Post a Comment