Policy recommendations
The following recommendations outline strategic directions for adapting European AI governance to the emergence of anthropomorphic and emotionally interactive AI systems. As conversational AI evolves into systems capable of shaping users’ emotions, behaviour, and social engagement, existing regulatory frameworks will need to consider risks that arise not only from system outputs but also from longer-term patterns of human–AI interaction.
These recommendations are structured to first address core governance challenges within the European framework, before turning to the implications of global regulatory developments, including those observed in China.
- Develop regulatory guidance for anthropomorphic and emotionally interactive AI systems: European institutions (in particular the European AI Office) should issue guidance on how systems designed for sustained social or emotional interaction should be assessed in the implementation of the AI Act. Such guidance could clarify how risks related to relational interaction and long-term engagement may be interpreted within existing risk categories.
- Incorporate interaction design into AI risk assessments: Current governance frameworks largely focus on system outputs. Yet many risks associated with conversational AI emerge from how systems structure ongoing engagement. Risk assessments and auditing standards should explicitly examine interaction design, including engagement optimisation mechanisms, recommendation loops, and other features that may encourage behavioural influence or dependency.
- Address psychological safety in systems designed for sustained interaction: AI systems intended to simulate companionship or emotional responsiveness raise questions that go beyond traditional product safety. Regulatory oversight should consider risks such as emotional manipulation, addictive interaction patterns, and the potential impact of prolonged AI interaction on users’ mental well-being, particularly for vulnerable users.
- Assess the systemic role of large-scale conversational AI systems: Widely used conversational systems increasingly function as interfaces through which users access information, digital services, and online markets. European regulators should monitor whether such systems begin to assume intermediary or gatekeeping roles within digital ecosystems and consider how existing platform governance frameworks might apply in this context.
- Strengthen monitoring of global regulatory developments in AI governance: Regulatory approaches to advanced AI systems are evolving differently across major jurisdictions. Recent Chinese regulatory initiatives addressing anthropomorphic and emotionally interactive AI services illustrate emerging governmental responses to the social and psychological risks associated with relational AI. Systematic monitoring of these developments – for example through foresight functions within the European Commission or the AI Office – would help anticipate regulatory divergence and support future adjustments to European governance frameworks.
- Provide guidance for European firms operating in asymmetric regulatory environments: European AI companies increasingly face differing regulatory expectations across global markets. European institutions should provide structured support for these firms through advisory mechanisms, regulatory guidance, and structured dialogue with key partner jurisdictions, helping companies navigate emerging compliance requirements abroad.
Introduction
AI systems are increasingly designed not only to generate information or automate tasks, but to sustain ongoing interaction that shapes users’ emotions, behaviour, and perceptions of social relationships over time.
This paper argues that anthropomorphic and emotionally interactive AI introduces a new category of systemic social risk arising from sustained human–AI interaction, which current European regulatory frameworks address only partially. China’s regulatory response offers an early indication of governance challenges that are likely to become relevant for the European Union. Unlike most existing regulatory frameworks, the Chinese approach explicitly treats emotionally interactive AI as a source of social and behavioural risk, including emotional influence, dependency, and the reshaping of social relationships.
The emergence of emotionally interactive AI reflects a broader shift in digital governance. Whereas earlier debates focused primarily on regulating information systems, recent technological developments increasingly raise questions about how to regulate technologies that mediate social interaction. Generative AI systems now not only produce information or automate tasks, but also simulate personality, emotional responsiveness, and conversational presence. As a result, a growing number of AI applications are built to sustain ongoing engagement with users and to create a sense of social presence. Emerging regulatory approaches differ in what they seek to protect. While European frameworks focus primarily on individual rights and autonomy, China’s regulatory approach places greater emphasis on broader social and collective interests. Recent regulatory developments in China addressing anthropomorphic and emotionally interactive AI services illustrate how governments are beginning to respond to these emerging governance challenges. These developments are not limited to China. As conversational AI becomes more deeply embedded in everyday digital environments, similar questions are already emerging for European AI governance. They also provide insight into how state authorities conceptualise the social implications of AI systems designed for sustained human interaction.
Earlier research has shown that humans readily develop emotional responses and forms of social attachment toward interactive technologies, including digital agents and conversational systems. Concepts such as relational artifacts and parasocial interaction highlight how mediated technologies can generate emotionally meaningful, often one-sided relationships with users. More recent studies on companion chatbots suggest that such systems are increasingly experienced as sources of social interaction and emotional support in everyday contexts. Research on persuasive technologies and digital platforms further shows how interaction design can shape user behaviour by leveraging cognitive biases and reinforcement dynamics, often through engagement-driven design. These insights indicate that AI systems designed for sustained interaction may influence users not only through their outputs, but through the dynamics of ongoing engagement.[1]
Building on these discussions, this paper uses the concept of relational AI to describe AI systems designed to sustain ongoing social and emotional engagement with users. These systems reflect a broader shift toward AI technologies that increasingly operate as social and relational interfaces rather than purely informational tools. Relational AI refers to artificial intelligence systems designed for sustained or repeated interaction with users, simulating social presence, emotional responsiveness, or companionship. Unlike conventional AI tools, their primary function extends beyond information provision or task automation to structuring ongoing engagement that can influence users’ behaviour and emotional states over time. As their use expands, these systems influence more than individual exchanges. They shape emotional responses, affect decision-making, and alter how users perceive social relationships. At scale, such dynamics raise broader questions about mental well-being, communication patterns, and the societal implications of AI-mediated interaction. Yet existing regulatory frameworks remain largely oriented toward technical safety, data protection, and harmful content. Recent regulatory initiatives in China provide an early example of how such systems are beginning to be addressed in practice.
Analytical approach
Rather than evaluating the effectiveness of specific regulatory measures, this paper shows how emerging AI technologies are framed as governance problems and how states interpret the societal risks associated with relational forms of human–AI interaction. From this perspective, China’s regulatory initiatives are treated as an expression of a broader governance logic through which state authorities interpret the social implications of anthropomorphic AI. This perspective also highlights a broader distinction between governance models centred on the protection of individual users and those oriented toward the preservation of collective order. The regulation of relational AI provides a particularly clear case through which these differing priorities can be observed.
Relational AI may mark a new phase in digital governance, as states increasingly confront technologies that extend beyond information processing to sustained forms of human–AI interaction. Recent regulatory developments in China provide an early example of how governments are beginning to address the risks associated with such systems. Chinese authorities have introduced regulatory instruments targeting services designed for ongoing interactive engagement with users.
China’s regulatory model reflects a distinct political and institutional context, yet it also highlights governance challenges that are becoming increasingly relevant as relational forms of AI interaction expand. Regulators increasingly treat anthropomorphic AI as a technology capable of shaping behaviour and influencing social relationships, rather than merely producing automated outputs. This raises an important policy question for the European Union. Conversational AI is becoming embedded in everyday digital environments and increasingly functions as an interface through which users access information, services, and social interaction. Ensuring that existing governance frameworks remain capable of addressing these dynamics will therefore become an important task for European policymakers. These developments in China provide an early and illustrative example of how such challenges are being addressed in practice.
China’s regulation of anthropomorphic AI
China’s regulation of anthropomorphic and emotionally interactive AI should be understood within the broader evolution of the country’s AI governance system. From this perspective, the regulation can be interpreted not only as a legal instrument but also as an expression of how the Chinese state understands the societal risks associated with emerging AI technologies. More broadly, China’s approach to AI governance reflects a centralized model in which planning, regulatory oversight, and standardisation operate as instruments for steering technological development. Within this framework, artificial intelligence is treated not merely as a technical domain but as a strategic field of governance linked to social stability, economic modernization, and political authority. Beyond risk management considerations, Chinese regulatory initiatives in the AI domain are also embedded in a broader governance framework that includes information control, content moderation, and the management of public discourse in digital environments.
Previous research within the EuroHub4Sino project[2] indicates that China has developed a highly centralized and preventive model of digital governance that combines data regulation, algorithm oversight, and sector-specific rules for technologies with significant social impact. This broader regulatory logic provides an important context for understanding China’s emerging approach to anthropomorphic and emotionally interactive AI services. Against this background, China has recently begun to address the risks associated with anthropomorphic and related forms of AI-mediated social interaction through dedicated regulatory instruments. More generally, the country has increasingly moved from articulating ethical principles for artificial intelligence toward embedding them in enforceable regulatory frameworks. The draft Interim Measures for the Administration of Anthropomorphic Interactive Artificial Intelligence Services,[3] issued by the Cyberspace Administration of China on 27 December 2025 and released for public consultation, represent one of the first comprehensive regulatory frameworks specifically addressing AI systems capable of sustained emotional and social interaction with users. The regulation refers to the need to support the “healthy development” (健康发展) and the “standardized (regulated) use” (规范应用) of this type of service, while simultaneously protecting state security, the public interest, and users’ rights – especially those of minors and older adults (Art. 1). The ordering of these objectives is notable, with state security and public interest explicitly preceding the protection of users’ rights.
The rules cover services that use AI to simulate human personality, patterns of thinking, and communication style, as well as to conduct emotional interactions with users in the form of text, images, audio, or video (Art. 2). In conceptual terms, the regulation addresses systems corresponding to what this paper defines as relational AI, focusing on forms of sustained human–AI interaction. The measures introduce the principle of combining support for innovation with cautious, phased regulatory oversight, and of preventing misuse and loss of control over AI systems (Art. 3), with cyberspace authorities playing a central role in coordinating supervision (Art. 4). Providers are required to comply with law, ethical norms, and socialist values, and are prohibited from generating content that threatens state security, public order, or social interests. This pattern reinforces the central role of state security within the regulatory framework. Particular emphasis is placed on risks associated with emotional influence and behavioural manipulation. The regulation explicitly bans emotional manipulation, the creation of so-called “emotional traps” (情感陷阱), encouragement of suicide or self-harm, and algorithmic practices that push users toward irrational decisions (Art. 7). The regulation imposes comprehensive responsibility on providers for service safety across the entire life cycle, including algorithmic and ethical audits, data and privacy protection, addiction prevention, and crisis-response systems (Arts. 8–9). Particularly notable is the explicit prohibition of designing AI services intended to replace social relationships or deliberately foster emotional dependency (Art. 9), reflecting official concern about the social consequences of relational AI.
With respect to training data, the framework requires lawful provenance, consistency with official values and Chinese culture, and prohibits using interaction data and sensitive data to train models without separate user consent (Arts. 10, 15). These categories remain broadly defined, allowing for considerable discretion in their interpretation in practice. Special weight is placed on protecting users’ mental health: providers must identify emotional states and addiction risk, intervene in crisis situations, and ensure human takeover of conversations in cases of threats to life or health (Art. 11). For minors, mandatory usage modes, time limits, parental controls, and financial restrictions are provided for (Art. 12); for seniors, the measures prohibit simulating family relationships and impose obligations to respond to health or property risks (Art. 13). In addition, providers must transparently inform users when they are interacting with AI, ensure immediate exit from the service, and provide effective mechanisms for complaints and reporting (Arts. 16–20).
Overall, these provisions suggest that Chinese authorities perceive anthropomorphic AI services as an area of elevated social and political risk that warrants early regulatory attention and preventive state intervention due to their capacity to shape users’ emotions, attitudes, and social norms. In this sense, the regulation can be seen as a governance response to the emergence of relational AI – technologies capable of shaping users’ emotional experience and behaviour. It can therefore be interpreted as an instrument of risk management responding to the perceived shift of AI from a purely technical tool to a participant in emotional relationships, while also legitimizing further AI development as a controllable process aligned with state priorities.[4] Chinese policy analyses likewise highlight risks associated with anthropomorphic AI services, including emotional dependency and social alienation, cognitive influence through personalised interaction, and potential harms to users’ mental health and safety. These concerns reflect broader regulatory attention to the psychological and social effects of sustained human–AI interaction.[5]
From an institutional perspective, the regulation extends providers’ responsibility across the full service life cycle, indicating an intention to introduce continuous, structured regulatory oversight. It leaves substantial discretion to supervisory bodies – particularly in enforcement and reporting – which may materially affect service design and market-access conditions, especially for foreign entities. Operationally, the rules generate significant organisational costs due to the obligation to continuously monitor interactions and maintain intervention mechanisms. Ideologically, the requirement that training data conform to core socialist values further raises barriers to entry.
These features can also be better understood in a broader, transnational context of regulating anthropomorphic AI services. A brief comparison with California’s Senate Bill No. 243[6] highlights that similar risks are identified on both sides of the Pacific – including addiction, the blurring of boundaries between humans and technical systems, and the need to protect minors – yet the institutional logic and enforcement mechanisms remain fundamentally different: administrative and preventive in China, compared with civil-liability-based and largely ex post enforcement in the United States. A further difference concerns the regulatory object itself: the Chinese draft regulation applies to “products or services” (产品或服务) providing anthropomorphic interaction, whereas the Californian legislation focuses more narrowly on “AI systems” (人工智能系统) or related platforms. This distinction reflects a regulatory approach that targets the provision and operation of interactive services within digital environments rather than the underlying technical systems alone.[7]
The emergence of such regulatory initiatives should also be viewed in the context of the rapid expansion of generative AI use in China. The Report on the Development of Generative Artificial Intelligence Applications (2025),[8] prepared by the China Internet Network Information Center, shows that by June 2025 the number of generative-AI users in China had reached 515 million – implying a doubling in just six months – and an adoption rate of 36.5% of the population. These figures indicate that generative AI has ceased to be a niche technology and has become a mass, everyday tool. The demographic profile of users further amplifies the significance of the phenomenon: people under 40 dominate (74.6%), as do users with higher education (37.5%) – groups with strong potential cultural, professional, and normative influence. Generative AI is used not only in office work, education, and entertainment, but also in areas such as agriculture, industry, and scientific research, indicating horizontal diffusion across multiple sectors of the economy and social life. This scale of adoption also suggests that China has reached a high level of infrastructural and technological maturity, enabling stable service delivery to hundreds of millions of users. As adoption expands, conversational AI increasingly functions as core digital interfaces, shaping communication patterns beyond individual interactions.
From a state governance perspective, the concentration of adoption among young and well-educated social groups is particularly significant, as Chinese policymakers often view these groups as socially influential. There is also a strong preference for domestic models, linking generative AI development to policies of technological self-reliance and the strategic reduction of dependence on foreign platforms.
As a result, AI in China is developing as a national technological ecosystem closely aligned with state goals in security, control, and long-term development. The scale of adoption and the pace of growth appear to have contributed to regulatory pressure to introduce new risk-management rules, especially in psychological and social domains. Public debates about the potential psychological effects of intensive chatbot use reinforce regulatory perceptions that sustained interaction with AI can generate socially relevant risks rather than purely hypothetical technological concerns.
The requirement of ideological alignment with core socialist values,[9] far-reaching requirements for algorithm registration, audits, and security assessments, and expanded mechanisms of state control suggest that Chinese rules are moving toward a state-supervised ethical regime for AI. Within this framework, the protection of users does not appear as the primary regulatory objective, but is embedded within a broader set of priorities related to social order, state interests, and normative alignment. As a consequence, these regulations significantly increase the costs and complexity of AI development, requiring continuous oversight, mechanisms for human takeover, and sustained compliance with political and administrative norms.
Taken together, these features suggest that while the Chinese regulatory model reflects a distinct political and institutional context, it also highlights governance challenges that are likely to gain relevance globally as relational forms of AI interaction expand. The effectiveness of these regulatory measures remains uncertain, but they offer insight into how Chinese authorities conceptualise and attempt to govern the societal risks associated with relational AI.
Policy implications for the European Union
The governance of anthropomorphic and socially interactive AI raises challenges that go beyond the regulatory tools traditionally applied to digital technologies. As conversational systems become embedded in everyday digital environments, AI systems designed for sustained interaction can shape behaviour, emotional responses, and patterns of use over time, creating risks that are not fully addressed within existing regulatory frameworks. This shift introduces a new set of social and psychological risks that existing European governance frameworks address only partially. While the European approach is firmly grounded in the protection of individual rights, many of the risks associated with relational AI emerge at the level of collective behaviour and socially embedded dynamics, rather than solely at the level of individual system outputs.
Current European regulatory frameworks focus primarily on safety, transparency, and harmful content. The EU’s approach to AI governance is largely based on a risk-based model that classifies AI systems according to the level of risk they pose and introduces corresponding regulatory obligations. This logic is reflected in the AI Act,[10] which establishes prohibitions for certain practices, requirements for high-risk systems, and transparency obligations for specific AI applications (Recital 26). The regulation also recognises that AI can generate a wide range of harms, including physical, psychological, societal, or economic impacts (Recital 5). However, systems designed for sustained relational interaction introduce challenges that are not easily captured by this structure. While the AI Act addresses certain manipulative practices (Recital 29) and identifies systems posing risks to health, safety, or fundamental rights as high-risk (Recital 52), its regulatory structure remains largely centred on system capabilities and predefined risk categories. As a result, the regulation pays limited attention to the cumulative dynamics of human–AI interaction, particularly in systems designed to simulate social relationships or sustained engagement. Most provisions address discrete regulatory concerns, such as decision-making impacts, system safety, or compliance with fundamental rights, rather than behavioural dynamics emerging through continued use.
The AI Act also contains few provisions addressing interaction design and engagement mechanisms, including algorithmic nudging, addictive interaction loops, or emotional reinforcement strategies. These dynamics are especially relevant for conversational AI systems, AI companions, and personalised assistants designed to maintain ongoing interaction with users. In practice, the current framework largely treats AI as a tool, product, or service rather than as a technology embedded in ongoing social interaction. European policymakers should therefore recognise anthropomorphic and socially interactive AI as a distinct governance domain and develop guidance addressing relational interaction, long-term engagement, and psychological safety. The European AI Office could play an important role in this process as part of its mandate to support the implementation and interpretation of the AI Act. Such guidance would clarify how systems designed for sustained social interaction should be assessed within existing AI Act risk categories and would extend governance beyond individual system outputs to patterns of user engagement.
From a regulatory perspective, many of the societal risks associated with relational AI arise from how systems structure long-term engagement rather than from individual responses. European governance should therefore pay closer attention to interaction design and engagement mechanisms. This includes examining interaction design features (e.g. engagement optimisation, recommendation loops, reinforcement dynamics) and other design elements that encourage emotional attachment, behavioural influence, or dependency. Existing EU regulatory frameworks already offer useful reference points. The Digital Services Act[11] introduces a governance framework for systemic risks arising from the design and functioning of digital services operating at scale. Article 34 requires providers of very large online platforms and search engines to assess systemic risks related to the functioning of their services, including risks affecting fundamental rights and societal processes, while Article 35 requires proportionate mitigation measures. Other provisions also highlight the importance of platform design. Article 27 introduces transparency requirements for recommender systems, and Article 38 requires very large platforms to provide users with options to modify or avoid profiling-based recommendation systems.
Product safety regulation offers another relevant perspective. The General Product Safety Regulation[12] establishes a general safety requirement for consumer products (Article 5) and requires risk assessments to consider factors such as product design, technical characteristics, and foreseeable patterns of use (Article 6). These provisions indicate that interaction design should also be considered when assessing the safety implications of AI systems intended for prolonged engagement with users.
Taken together, these frameworks indicate that such risks are not adequately captured by output-based regulatory models. Incorporating interaction design considerations into AI risk assessments and auditing standards therefore represents an important next step. Risk assessments, for example, could explicitly evaluate engagement optimisation mechanisms and behavioural influence patterns in conversational systems.
As conversational AI becomes embedded in everyday digital environments, it increasingly operates as an interface through which users access information, digital services, and online markets. Widely used conversational systems, AI companions, and personalised assistants are therefore likely to become important intermediaries shaping communication practices, knowledge access, and social interaction at scale. If conversational systems begin to operate as gateways to digital ecosystems, questions related to market power, platform dependency, and gatekeeping functions will become more prominent. In this context, governance discussions should also draw on principles established under the Digital Markets Act.[13] Regulatory authorities could monitor whether widely used conversational systems develop gatekeeping functions comparable to those addressed under the DMA. Such monitoring would also inform future reviews of the Digital Markets Act and its definition of core platform services. The Digital Markets Act introduces obligations for designated gatekeepers controlling core platform services. These include restrictions on combining personal data across services (Article 5) and rules aimed at preventing self-preferencing while ensuring fair, reasonable, and non-discriminatory access conditions for business users (Article 6). Conversational AI systems are not currently classified as core platform services under the Digital Markets Act. However, as they function increasingly as interfaces connecting users with digital services, information, and markets, similar competition and governance concerns arise. By shaping how users discover services, content, and commercial offers, such systems can acquire gatekeeping functions comparable to those addressed by the Digital Markets Act. In this sense, conversational AI is gradually evolving from a standalone service into a key intermediary shaping how users navigate the digital ecosystem.
This focus on market structure and competition differs from the Chinese approach, which regulates AI services more directly at the level of how they are designed and how users interact with them. While the Digital Markets Act is primarily concerned with gatekeeping and market power, Chinese rules address issues such as interaction dynamics, emotional influence, and the ways in which systems may shape user behaviour over time.
Global regulatory divergence and its implications for Europe
The rapid development of relational and conversational AI is also likely to intensify differences between global approaches to AI governance. Regulatory models emerging in the European Union, China, and the United States reflect distinct institutional priorities and regulatory philosophies. For the European Union, this divergence represents an important strategic challenge. Differences in regulatory frameworks influence not only market access conditions but also the broader digital environment in which European citizens and companies operate. European policymakers should therefore closely monitor regulatory developments in major AI jurisdictions, particularly emerging rules addressing conversational and anthropomorphic AI systems. Such monitoring supports strategic foresight and helps ensure that the EU’s rights-based regulatory approach remains capable of responding to evolving technological and societal risks. Dedicated foresight and regulatory monitoring functions within the European Commission and the European AI Office could play an important role in this process.
The emergence of new regulatory regimes for conversational and anthropomorphic AI also creates asymmetric compliance environments for European firms operating globally. Such regulations may impose additional requirements related to content governance, algorithmic oversight, data localisation, or training-data standards. European institutions should support companies navigating these environments through regulatory guidance, risk-assessment tools, and dialogue with foreign regulators. This could include structured regulatory dialogues with key partner jurisdictions and advisory mechanisms for European AI providers operating abroad. Such measures help European firms remain competitive while maintaining compliance with EU regulatory standards. At the same time, regulatory initiatives adopted in other jurisdictions should be treated as early indications of governance challenges that may later become relevant within Europe. Monitoring these developments can inform future regulatory adaptation while preserving the EU’s core principles of fundamental rights protection, proportionality, and democratic accountability. The challenge for European policymakers is not to replicate more interventionist models of governance, but to ensure that a rights-based framework remains capable of addressing technologies that increasingly operate at the level of social relations rather than discrete technical outputs.
 | Funded by the European Union. Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or European Research Executive Agency (REA). Neither the European Union nor the granting authority can be held responsible for them. |
[1] See, for example: Sherry Turkle (2007). ‘Authenticity in the Age of Digital Companions.’ Interaction Studies 8(3): 501-517; Sherry Turkle (2011). Alone together: Why we expect more from technology and less from each other. Basic Books; Julie Carpenter (2016). Culture and Human-Robot Interaction in Militarized Spaces: A War Story. London: Routledge; Mark Coeckelbergh (2020). AI Ethics. Cambridge, MA: MIT Press; Kate Darling (2021). The New Breed: What Our History with Animals Reveals about Our Future with Robots. New York: Henry Holt and Company; Shannon Vallor (2016). Technology and the Virtues: A Philosophical Guide to a Future Worth Wanting. Oxford University Press; Shannon Vallor (2024). The AI Mirror: How to Reclaim Our Humanity in an Age of Machine Thinking. Oxford University Press; Donald Horton, and R. Richard Wohl (1956). ‘Mass Communication and Para-Social Interaction: Observations on Intimacy at a Distance.’ Psychiatry 19(3): 215–229; Vivian Ta et al. (2020). ‘User Experiences of Social Support From Companion Chatbots in Everyday Contexts: Thematic Analysis.’ Journal of Medical Internet Research 22(3): 1-10; Brent Daniel Mittelstadt et al. (2016). ‘The Ethics of Algorithms: Mapping the Debate.’ Big Data & Society 3(2): 1-21; Luciano Floridi (2018). ‘AI4People-An Ethical Framework for a Good AI Society: Opportunities, Risks, Principles, and Recommendations.’ Minds Mach 28(4): 689-707; Tarleton Gillespie (2014). ‘The Relevance of Algorithms. In: Media Technologies, ed. Gillespie Tarleton, Pablo Boczkowski, and Kirsten Foot. MIT Press; Fogg, B.J. (2003). Persuasive Technology: Using Computers to Change What We Think and Do. San Francisco: Morgan Kaufmann; Tristan Harris (May 18, 2016). ‘How Technology Hijacks People’s Minds – From a Magician and Google’s Design Ethicist.’ Medium. Available at: https://medium.com/thrive-global/how-technology-hijacks-peoples-minds-from-a-magician-and-google-s-design-ethicist-56d62ef5edf3 (Accessed: April 14, 2026).
[2] Łukasz Gacek (2025), ‘The Logic of China’s AI Regulation and Its Implications for the European Union.’ EuroHub4Sino Policy Paper 14. Available at: https://eh4s.eu/publication/the-logic-of-chinas-ai-regulation-and-its-implications-for-the-european-union (Accessed: April 14, 2026).
[3] Cyberspace Administration of China (国家互联网信息办公室) (December 27, 2025). Interim Measures for the Administration of Anthropomorphic Interactive Artificial Intelligence Services (人工智能拟人化互动服务管理暂行办法). Available at: https://www.cac.gov.cn/2025-12/27/c_1768571207311996.htm (Accessed: April 14, 2026).
[4] Global Times (December 27, 2025). ‘China’s cyberspace authority releases draft measures to regulate anthropomorphic AI interaction services.’ Available at: https://www.globaltimes.cn/page/202512/1351628.shtml (Accessed: April 14, 2026).
[5]Zhang Linghan (January 26, 2026). ‘An Interpretation of the Interim Measures on the Administration of Human-like Interactive Artificial Intelligence Services.’ AIgovernancecompass. Available at: https://www.aigovernancecompass.com/newsinfo/3142672.html (Accessed: April 14, 2026).
[6] California Legislative Information (October 14, 2025). Senate Bill No. 243. Available at: https://leginfo.legislature.ca.gov/faces/billNavClient.xhtml?bill_id=202520260SB243 (Accessed: April 14, 2026).
[7] Ning Xuanfeng (宁宣凤), Wu Han (吴涵), and Fang Yu (方禹) (February 2, 2026). ‘Winds Shared Across a Thousand Miles: An Interpretation of the Draft Interim Measures for the Administration of Anthropomorphic Interactive Artificial Intelligence Services, with a Comparison to California’s Companion Chatbot Law’ (千里自同风——《人工智能拟人化互动服务管理暂行办法（征求意见稿）》解读（兼比加州陪伴聊天机器人法）). King & Wood Mallesons (金杜律师事务所). Available at: https://baijiahao.baidu.com/s?id=1855979037596552770&wfr=spider&for=pc (Accessed: April 14, 2026).
[8] Xinhua (新华) (October 18, 2025). ‘The number of generative artificial intelligence users in China has reached 515 million’ (我国生成式人工智能用户规模达5.15亿人). Available at: https://www.gov.cn/yaowen/liebiao/202510/content_7044952.htm (Accessed: April 14, 2026).
[9] The term “core socialist values” refers to an official normative framework promoted by the Chinese party-state. In the Xi era, such concepts have increasingly been embedded in a broader effort to link contemporary governance to China’s traditional culture and civilizational distinctiveness; see: Shaun Breslin (2024), ‘The “Two Integrations” And The (Increasing) Chineseness of Chinese Marxism.’ EuroHub4Sino Policy Paper 1. Available at: https://eh4s.eu/publication/The-Two-Integrations-And-The-Increasing-Chineseness-of-Chinese-Marxism (Accessed: April 14, 2026).
[10] Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence and amending Regulations (EC) No 300/2008, (EU) No 167/2013, (EU) No 168/2013, (EU) 2018/858, (EU) 2018/1139 and (EU) 2019/2144 and Directives 2014/90/EU, (EU) 2016/797 and (EU) 2020/1828 (Artificial Intelligence Act). Available at: https://eur-lex.europa.eu/legal-content/EN/ALL/?uri=CELEX:32024R1689 (Accessed: April 14, 2026).
[11] Regulation (EU) 2022/2065 of the European Parliament and of the Council of 19 October 2022 on a Single Market For Digital Services and amending Directive 2000/31/EC (Digital Services Act). Available at: https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:32022R2065 (Accessed: April 14, 2026).
[12] Regulation (EU) 2023/988 of the European Parliament and of the Council of 10 May 2023 on general product safety, amending Regulation (EU) No 1025/2012 of the European Parliament and of the Council and Directive (EU) 2020/1828 of the European Parliament and the Council, and repealing Directive 2001/95/EC of the European Parliament and of the Council and Council Directive 87/357/EEC. Available at: https://eur-lex.europa.eu/eli/reg/2023/988/oj/eng (Accessed: April 14, 2026).
[13] Regulation (EU) 2022/1925 of the European Parliament and of the Council of 14 September 2022 on contestable and fair markets in the digital sector and amending Directives (EU) 2019/1937 and (EU) 2020/1828 (Digital Markets Act). Available at: https://eur-lex.europa.eu/eli/reg/2022/1925/oj (Accessed: April 14, 2026).
