Shlomit Wagman, a former Israeli financial regulator, suggests that the US and China can establish AI safety agreements based on mutual self-interest rather than political trust. She argues that narrow pacts focusing on existential risks, such as loss-of-control scenarios and bioweapons development, are more viable than seeking comprehensive geopolitical alignment.
As of September 2026, US-China dialogues on AI risks have been active, focusing on shared vulnerabilities and crisis management. These discussions prioritize practical verification tools and hotlines over broad consensus. This approach aligns with a September 2026 analysis in Foreign Policy, which emphasizes the necessity of technical safeguards over declarations of shared values.
The urgency for such frameworks is highlighted by a November 2025 report from Anthropic, which detailed an AI-enabled espionage campaign against US infrastructure attributed to Chinese state actors. Furthermore, researchers at a recent closed meeting in San Francisco warned that AI capabilities may be advancing faster than the current ability to control them.
Wagman proposes that agreement structures should be designed to make non-compliance less attractive than adherence. For the tech sector, this shift toward government-backed safety frameworks is expected to increase funding for defensive AI systems, governance tools, and verification technologies, potentially boosting demand for companies specializing in risk management.
