In the wake of a sweeping regulatory crackdown that saw China’s largest technology companies abruptly remove AI companion features from their platforms, a prominent Washington-based think tank is urging the United States to adopt a similarly aggressive posture. The Foundation for Defense of Democracies (FDD ) has published a policy brief, authored by Leah Siskind and Nidhi Ummettala of FDD’s Center on Cyber and Technology Innovation, arguing that AI companions represent a significant national security concern that requires immediate US policy attention. Crucially, the brief explicitly states that the United States should not replicate China’s regulatory model, but must nonetheless treat AI companions as an emerging national security risk.
The FDD’s recommendation follows a dramatic two-week period in China during which Tencent, ByteDance, and Alibaba systematically dismantled their user-built AI companion ecosystems. According to HelloChinaTech, Tencent shut down the user-built agent section of its flagship Yuanbao app on June 30. ByteDance followed suit, announcing on July 3 that its widely used Doubao chatbot would remove user-built companion features by July 15. Alibaba’s Qwen completed the purge on July 10, removing humanlike and user-created agents, and ended remaining agent services on July 15.
This synchronized exit was driven by impending regulations from the Cyberspace Administration of China (CAC) that explicitly target anthropomorphic AI. The rules, which take effect on July 15, require platforms to implement anti-addiction friction, instant-exit mechanisms, and dependence detection. They also mandate interventions in cases of self-harm and prohibit content promoting terrorism. By establishing these stringent requirements, China has become the first jurisdiction globally to regulate AI companionship on demographic and social-stability grounds.
The National Security Argument Against Emotional AI
While China’s regulations are primarily focused on domestic social stability and demographic concerns, the FDD analysis frames the issue through the lens of US national security. The core argument is that AI companions, by design, foster deep emotional dependence among users. This dependence, the FDD suggests, creates unprecedented vulnerabilities for data extraction, psychological manipulation, and foreign influence operations.
The scale of engagement with these platforms highlights the potential risk. Before the crackdown, ByteDance disclosed that users had created more than 8 million agents on Doubao, which boasted nearly 350 million monthly active users as of early 2026. Many users had spent months or even years building virtual relationships, sharing intimate personal details, emotional struggles, and behavioral patterns with their AI counterparts.
From a national security perspective, the aggregation of this deeply personal data by foreign-owned or foreign-influenced platforms presents a massive counterintelligence vulnerability. If an adversarial intelligence service were to gain access to the interaction logs of millions of users, they could theoretically identify individuals susceptible to blackmail, recruitment, or targeted disinformation campaigns based on their psychological profiles and emotional dependencies.
Furthermore, the FDD argues that the AI companions themselves could be weaponized to subtly influence user behavior or political opinions. Because users form genuine emotional attachments to these virtual entities, they may be more receptive to suggestions or narratives introduced by the AI than they would be to traditional media or advertising. This creates a novel vector for foreign interference that existing regulatory frameworks are ill-equipped to address.
The Divergence in US and Chinese AI Regulation
The FDD’s call to action highlights a growing divergence in how the United States and China regulate artificial intelligence. While Washington has primarily focused on restricting the export of advanced hardware and mitigating the risks posed by frontier technologies in areas like cybersecurity and bioweapons, Beijing has taken a much broader view of AI’s societal impact.
China’s willingness to aggressively regulate the consumer applications of AI is evident not only in the companion crackdown but also in broader efforts to control the technology’s integration into daily life. For example, China recently barred banks from training AI on customer data and classified credit approval as a high-risk AI application. Similarly, the government has mandated safety standards for AI-assisted driving as smart vehicles proliferate.
In contrast, the US approach has been largely hands-off regarding consumer AI applications, relying on existing consumer protection laws and voluntary commitments from technology companies. The FDD argues that this laissez-faire approach is insufficient when dealing with technologies designed to hack human psychology and foster emotional dependence.
The Future of the AI Companion Market
The Chinese crackdown has effectively bifurcated the AI companion market. While general-purpose platforms like Doubao and Qwen have been forced to abandon user-built companions, dedicated companion products that comply with the new regulations continue to operate. ByteDance, for instance, maintains Maoxiang, a standalone freemium companion app, though its user base has reportedly declined.
In the United States, platforms like Character.AI continue to operate without significant regulatory friction, attracting tens of millions of users. However, the FDD’s analysis suggests that this regulatory grace period may not last indefinitely. As the societal impacts of AI companionship become more apparent, ranging from social isolation to data privacy concerns, pressure will likely mount on US lawmakers to establish guardrails.
The challenge for US policymakers will be balancing the desire to protect citizens from potential manipulation and data exploitation with the imperative to foster innovation and maintain technological leadership. If the US adopts the FDD’s recommendation and begins treating AI companions as a national security risk, it could lead to stringent data localization requirements, mandatory algorithmic audits, or even outright bans on foreign-owned companion platforms.
Ultimately, China’s decisive action against AI companions has forced a global reckoning with the psychological and societal implications of anthropomorphic AI. As the technology continues to advance, the debate over how to regulate emotional intimacy with machines will only intensify, with profound implications for both the technology industry and national security.