The episode explores the rise of “humanised” AI, including virtual companions designed to simulate emotional intelligence and human relationships. Our speakers examine what is driving the rapid adoption of these models, psychological risks of replacing human interactions and how regulatory approaches and user perception differ between Asia and Europe. The discussion highlights the need for ethical safeguards, multi-stakeholder collaboration, stronger social connections and social movements to address both the societal and private sector factors  fueling demand for digital companionship.  

The Straits Times Assistant Editor (Newsroom Strategy) Rei Kurohi hosts an international panel featuring Dr Gry Hasselbalch, (Co-founder DataEthics.eu); Ms Kristina Fong, (Fellow (Economics), ASEAN Studies Centre, ISEAS-Yusof Ishak Institute); and Assistant Prof (AP) Renwen Zhang (Director, SWEET Lab-Wee Kim Wee School of Communication and Information, NTU) to discuss the societal, ethical and policy implications of humanised AI. 

iiii iiii iiii iiii iiii iii iii i i iiii i  iiiii i  iiiii ii i iii ii i iiiiii   i iii iii i i i iii ii ii i ii iii i i ii i i iii iii ii ii ii ii ii 

With therapy and companionship emerging as top generative AI use cases in 2025 and 2026, conversational models are rapidly evolving from productivity utilities into socially and emotionally intelligent virtual companions.  With these always-on systems growing increasingly hyper-realistic, their adoption raises questions about human psychology, ethical AI design and regional governance.

1. The Drivers of the AI Companionship Surge

AP Zhang highlighted that while companion apps existed prior to the widespread adoption of large language models, the boom in generative AI transformed companionship from a niche subculture into a major market driver is projected to reach US$140 billion by 2030.  

“AI companions did not create loneliness, they just monetised it,” Asst Prof Zhang explained. She noted that the rise of these systems stems from a convergence of technological push for advanced models capable of simulating empathy and a pre-existing loneliness epidemic accelerated by societal shifts and the COVID-19 pandemic.  

However, the panel noted that these apps operated under rules that were fundamentally different from human relationships.  Because AI companions are designed to be friction-free, infinitely patient and agreeable, users are at risk of unintended side effects when interacting with AI that uncritically affirms desires and behaviors that a healthy human dynamic would challenge.  Over time, this raises concerns about the atrophy of essential social skills such as conflict resolution, compromise and perspective-taking, particularly among younger users.  Dr Hasselbalch also highlighted the potential impact in the context of democracy where friction and negotiation were core components.  

2. Divergent Regulatory Approaches: the European Union (EU) vs The Association of Southeast Asian Nations (ASEAN)

The discussion revealed sharp contrasts in how regions govern anthropomorphic AI.  Dr Hasselbalch detailed the European Union (EU)’s law-first approach, grounded in the EU AI Act.  She emphasised that true “human-centric” AI governance was not about making AI appear human, but about preserving human agency, enforcing transparency and curbing deceptive design patterns.  

“It’s about creating AI technologies with human agency at the center… transparent models, models that are non-deceptive by design,” Dr Hasselbalch noted, while cautioning that current European debates must balance these high fundamental-rights standards against competitive pressures to adopt “AI-first” strategies.  

In contrast, Ms Fong outlined ASEAN’s non-binding guidance framework.  Given the diverse economic development, digital readiness and institutional capacity across member states, ASEAN relies primarily on voluntary adherence to guidelines, regional best practices and domestic enforcement, rather than a centralised legal structure.  

She observed that current regional regulations focused heavily on deployment, system design and keeping a human in the loop, but rarely addressed the specific psychological and anthropomorphic impacts on end-users. She called for elevating these psychological and societal risks higher on public policy agendas.

3. User Resistance, Ethics and the Path to Cross-Regional Collaboration

Looking toward the future, the panel explored how society can maintain balance as humanised models continue to develop. Even as Big Tech interests “colonised” international forums on AI, Dr Hasselbalch noted the emergence of social movements and resistance, such as student pushback against the pervasive overuse of generative AI. She viewed this behaviour not as a rejection of the technology itself, but as a healthy democratic defense against influential tech companies operating unchecked by regulators.  She also advocated developing a “shared narrative” for a more coherent response.   

Ms Fong highlighted that while global conversations often focused on tangible governance, the ethical dimension is equally vital, leading major tech firms to increasingly recruit philosophers and ethicists to address complex human impacts.  

Closing the conversation, AP Zhang advocated for multi-stakeholder framework deliberations that bring diverse regions together to negotiate values, trade-offs and policy safeguards without inadvertently causing unintended secondary effects.  The panel agreed that addressing the challenge of humanised AI required going beyond regulatory constraints, and building societal conditions where people felt heard and connected in the real world.  

.    .    .    .    .