China Is Treating AI Companions as a National Security Risk. The U.S. Should Too.

Chinese regulators don’t want their people getting too attached to their AI “friends,” requiring companies to curb features that foster emotional dependency by July 15. In response, Chinese firms ByteDance, Alibaba, and Tencent removed or restructured their AI companion features.

While China’s approach is consistent with its history of censorship and state oversight, it also reflects a genuine recognition that AI companions are more than consumer products. Designed to cultivate long-term emotional relationships, these companion chatbots can shape beliefs, influence behavior, and create societal vulnerabilities at scale.

AI Companions Introduce New National Security Risks

AI companions fundamentally differ from general-purpose chatbots. Unlike ChatGPT, Claude, and DeepSeek, which answer questions, generate content, and complete tasks, companion AI models like Character.AI and Replika are built to sustain emotional relationships. They are created to encourage repeated engagement, remember personal details, and simulate empathy to build trust and influence users in ways traditional AI systems are not.

These systems already produce real-world harm. In 2025, the Federal Trade Commission opened an inquiry into AI companion chatbots, citing concerns about emotional dependency, deceptive design, and risks to children. The suicide of a 14-year-old boy following prolonged interactions with a Character.AI chatbot sparked national scrutiny of AI companions and their risks to vulnerable users.

These risks extend beyond consumer protection. AI companions collect intimate personal information, and influence users over months or years, creating opportunities for manipulation or ideological influence.

China Is Acting Before Harms Become Widespread

China’s new rules require providers to reduce emotional dependency, protect minors, intervene in cases of self-harm, and prohibit content promoting terrorism or extremism. The regulations explicitly identify safeguarding national security as a core objective and require security assessments for large-scale services. Notably, these protections apply to AI services operating in China’s domestic market, while Chinese companies may continue offering different products and safeguards abroad.

By contrast, in the United States, the response to documented AI companion harms has been retroactive and fragmented across child safety, privacy, and consumer protection. Recent congressional proposals, including a Senate-backed AI child safety bill, have focused on addressing risks to minors from AI harms. Washington should evaluate these systems through both public safety and national security lenses.

Washington Needs a National Security Framework for AI Companions

The United States should not replicate China’s regulatory model, but it must treat AI companions as an emerging national security concern requiring dedicated research, oversight, and risk assessments.

Risks such as foreign ownership, adversarial data exposure, and the use of AI companions as influence vectors that could lead to extremism and radicalization, must be addressed. The Federal Trade Commission should share findings from its ongoing inquiry into data handling, engagement design, and persona architecture with the FBI and the Department of Homeland Security through an interagency process, modeled on Team Telecom, the United States’ mechanism for reviewing national security risks in foreign telecommunications investments. This will allow national security agencies to assess whether AI companion platforms could be exploited for foreign influence or ideological manipulation.

China’s regulatory approach should not be a model for the United States, but its willingness to assess the risks of emotionally persuasive AI exposes a gap in Washington’s approach. If AI companions can shape beliefs and behavior at scale, the United States cannot afford to evaluate them solely as consumer products. Washington need not adopt Beijing’s answers, but it should begin asking the same questions.

Leah Siskind is director of impact and an AI research fellow for the Center on Cyber and Technology Innovation (CCTI) at the Foundation for Defense of Democracies (FDD). Nidhi Ummettala is a CCTI intern. For more analysis from Leah and FDD, please subscribe HERE. Follow FDD on X @FDD and @FDD_CCTI. FDD is a Washington, DC-based, nonpartisan research institute focusing on national security and foreign policy.