The discourse surrounding U.S.-China artificial intelligence relations has matured significantly, evolving beyond initial concerns regarding military applications to encompass frontier model risks, biosecurity, and cybersecurity. Yet, amidst this broadening agenda, the issue of disinformation remains strangely peripheral, treated primarily as a medium for mutual accusation rather than a shared governance challenge. As AI-generated content continues to permeate the digital landscape, the speed at which unverified media—such as deepfake videos or doctored screenshots—can propagate creates a dangerous instability. Increasingly, this volatility is outpacing diplomatic processes, forcing policymakers to react to fabricated crises before the underlying facts can be established or verified.
This challenge is exacerbated by a climate of profound mutual distrust between Washington and Beijing. In such high-stakes environments, the absence of instantaneous proof leads both nations to adopt a “worst-case scenario” interpretation of any ambiguous event, causing diplomatic tension to flare within hours of an incident. AI, in this sense, acts as a force multiplier for suspicion, effectively collapsing the period of reflection necessary for rational statecraft. Because the damage of disinformation often crystallizes well before the truth arrives, the two nations are frequently trapped in a cycle of reactive hostility, inadvertently conceding their policy autonomy to the viral speed of unverified content.
A critical lesson for managing this instability can be drawn from the Cold War-era “hotline” established between Washington and Moscow following the Cuban Missile Crisis. Although the direct communications link did not resolve the ideological or geopolitical disputes of the time, it provided essential breathing room during moments of international flashpoints, preventing hasty, irreversible escalations. By adopting a similar logic, the U.S. and China could establish a framework focused not on reconciling their competing narratives, but on creating the “temporal space” necessary for verification. This would prioritize crisis stability over the impossible goal of immediate, universal truth, helping both powers avoid political entrapment caused by digital deceptions.
Beyond the need for rapid communication, the structural challenges of AI disinformation are twofold: a technical gap and a machine-learning concern. Currently, the time required to generate convincing fakes is trivial, while the time required for professional forensic verification is protracted, creating an environment where public and political pressure makes “waiting for the facts” politically untenable. Simultaneously, there is a looming threat where low-quality, AI-generated disinformation is ingested by Large Language Models (LLMs) and cemented as “historical fact” in future AI training data. If unchecked, the current smog of disinformation will distort the foundational knowledge base upon which future AI assistants—and by extension, future human understanding of international relations—relies.
To address these vulnerabilities, the U.S. and China should pursue three pragmatic, non-idealistic steps. First, they should incorporate a “crisis protocol” into existing communication channels, allowing for direct consultation during suspicious incidents before public accusations are formalized. Second, they must prioritize the interoperability of content provenance standards, ensuring that AI-generated media remains traceable and verifiable even when it crosses national borders. Third, both governments should commit to making their authoritative records and primary documents available in machine-readable formats. By populating the internet with high-quality, verifiable primary sources, both nations can ensure that global LLMs have accurate data to draw upon, thereby inoculating the information space against the most egregious fabrications.
Ultimately, this proposal does not require the U.S. or China to abandon their systemic competition or set aside their fundamental grievances. Instead, it recognizes a narrow but vital shared interest: the preservation of a stable global information environment. If both nations allow the digital landscape to become a space where synthetic content dictates foreign policy, they both risk losing sovereignty over their own diplomatic responses. By building an infrastructure for verification, the two powers can reclaim the time necessary to form deliberate, evidence-based judgments, ensuring that, even in an era of intense rivalry, they are not governed by the malicious speed of a digital phantom.