From 2 August 2026, anyone offering a chatbot in the European Union (EU) must clearly tell users that they are interacting with a machine, not a person, under Article 50 of the EU AI Act. This may sound like a simple transparency rule for customer service bots or online help desks. Yet it comes at a time when chatbots are becoming far more personal. Humanised AI systems, designed to sound natural, conversational and emotionally responsive, are increasingly being used not only for information or productivity, but also for companionship, reassurance and emotional support. Research conducted by Marc Zao-Sanders for Harvard Business Review found that, in 2025, therapy and companionship have become the most popular individual uses of generative AI worldwide, while the broader category of ‘personal support’ has nearly doubled in a year to around 30 percent of all use. These tools can offer social value by providing a non-judgemental and constantly available space for people who feel lonely or anxious. However, they also pose risks of dysfunctional dependence and human interaction substitution. MIT sociologist Sherry Turkle warns that AI may offer only ‘simulated empathy’, creating the impression of care without real human understanding.
Asia and Europe are addressing this phenomenon, albeit with different approaches. In Europe, where autonomy, dignity and data protection are central concerns, the EU AI Act sets binding transparency requirements and penalties for non-compliance. In ASEAN, the expanded Guide on AI Governance and Ethics takes a more voluntary approach, identifying anthropomorphism as a key risk: the tendency of users to treat AI as if it truly understands, feels or behaves like a human. Both approaches respond to the same challenge: as AI becomes more personal, users need to know what they are interacting with, use it safely and protect their privacy and real human relationships.
Another key challenge is managing the underlying incentive – whether AI companions are designed mainly to keep people engaged, or to genuinely help them. The appeal of AI companions is real. Harvard Business School researchers found that talking with an AI companion eased loneliness about as much as talking with another person, largely because users felt heard. Japan, where close to four in 10 people have reported feeling lonely, has turned to AI companions partly as a response to demographic strain. Asia and Europe already have a concrete channel to compare notes on questions like these, notably the ASEAN-EU Digital Work Plan, where AI governance is one of the areas identified for deepening cooperation. Such channels will allow for discussions on the safeguards needed to ensure that humanised AI is transparent, accountable and has proper oversight.
Sources:
Joint Ministerial Statement 25th Asean-Eu Ministerial Meeting 27-28 April 2026
Shay Dalton, 2026, ‘How people are really using generative AI in 2025?’ Steering Point Advisory
European Commission, 11 May 2026, ‘AI Act’, European Commission
Expanded-ASEAN-Guide-on-AI-Governance-and-Ethics-Generative-AI.pdf
Post Your Comments