The relationship between the United States and China is defined by a complex web of technical and strategic dialogues, yet one glaring omission persists: a shared approach to the chaos of AI-driven disinformation. While leaders spend hours hashing out regulations on military AI, cybersecurity, and biosecurity, the issue of synthetic content remains trapped in a cycle of mutual accusation. This is a missed opportunity. We must pivot from using disinformation as a rhetorical weapon to treating it as a shared systemic threat. The danger is no longer just about the content itself, but about the alarming speed at which unverified information now circulates, often reaching global audiences and triggering diplomatic crises long before the underlying facts can be established. In a world of instantaneous digital contagion, truth is increasingly struggling to keep pace with the narrative.
This “speed trap” is creating a dangerous new normal for diplomacy. In today’s high-friction environment, missing information is rarely met with patience; it is met with the worst-case interpretation. When an unverified video claiming to show a provocation appears online, a government facing a massive, outraged public cannot simply wait for a formal investigation—the news cycle will have already moved on, and political pressure will force an immediate, potentially volatile response. AI has effectively collapsed the space between suspicion and conviction, compelling leaders to make life-altering policy decisions in mere hours. This is not just a challenge for journalists tethered to deadlines; it is a structural crisis that threatens the stability of international relations and heightens the risk of accidental escalation.
To find a way forward, we should look to the lessons of the Cold War. During the darkest days of the nuclear arms race, the U.S. and Soviet Union realized that they shared a common enemy in the form of catastrophic misunderstanding. The hotline established after the Cuban Missile Crisis did not resolve deep ideological disputes or end the Cold War, but it provided something far more practical: the gift of time. By allowing leaders a few precious hours to communicate, verify, and breathe, this link acted as a circuit breaker for potential disaster. Today, we need a similar, low-stakes mechanism for the information age. Cooperation doesn’t require us to agree on the truth of every issue, but it does require us to agree that both sides have a vested interest in avoiding a crisis sparked by a machine-generated lie.
The technical reality of this problem is stark: as it stands, we are fighting a losing battle. Creating a synthetic video now happens in minutes, while authoritative forensic verification can take days or weeks. Furthermore, the threat is evolving beyond human perception. As people increasingly rely on AI models and chatbots to synthesize information about global trade, geopolitics, and international relations, these models are essentially “digesting” the low-quality, polluted web to form their answers. If we don’t improve the quality of the information landscape today, we are essentially poisoning the well from which future generations will derive their understanding of reality. Labeling and taking down content are mere stopgaps; we need to build a more resilient information infrastructure that prioritizes the accessibility of primary, verifiable data.
Three modest, realistic steps could serve as a starting point for this cooperation. First, we must integrate “AI-incident” protocols into our existing crisis communication channels, giving diplomats a dedicated track to cross-check potentially inflammatory AI content before a formal government response is issued. Second, both nations should push for interoperability in content-provenance standards, ensuring that a digital “label” or signature placed on an AI-generated image in Washington can be read and decoded by a system in Beijing. This simple technical compatibility would allow both sides to track the origins of synthetic material. Finally, both governments ought to proactively release more official documentation in machine-readable formats. By feeding high-quality, truthful data into the global information pool, we ensure that AI models have a reliable source to draw from when they answer user questions.
Ultimately, these initiatives do not require a romantic vision of trust; they are rooted in the pragmatic, cold-eyed pursuit of stability. Structural competition between the U.S. and China is inevitable, and many of our disagreements will endure for years to come. However, the integrity of the global information environment is a shared public good. Neither capital benefits from a world where an AI-generated fiction can force a government’s hand before it can even authenticate what is happening. By focusing on the infrastructure of verification rather than ideological alignment, we can create a floor beneath our relationship—a necessary buffer that ensures we don’t accidentally walk into a conflict we never intended to start.