Singaporeans open to AI shopping agents but demand strict checks before payment

A Global Payments study found that 59 per cent of Singaporeans are open to using an AI agent for shopping, while only 15 per cent would let one browse, choose and pay independently. Security protections, certification and final approval remain central to wider adoption.

Editor: AI News 24x7 Published: 23 Sep 2026, 08:38 AM Updated: 23 Sep 2026, 08:38 AM

Singaporeans open to AI shopping agents but demand strict checks before payment.

Singaporeans are showing strong interest in artificial intelligence agents that can shop on their behalf, but most want firm safeguards before allowing the technology to complete a purchase. A Global Payments study found that curiosity about agentic AI is running ahead of confidence in fully autonomous buying.

The research covered 7 countries, including the United States, China and the United Kingdom. In Singapore, 59 per cent of respondents said they were open to an AI agent shopping for them, while 23 per cent had already used one. Only 15 per cent would allow an agent to browse, shortlist products and make payment entirely without human intervention.

Trust remains tied to human approval

The findings highlight a gap between using AI for recommendations and handing it responsibility for the final transaction. Potential uses include changing insurance policies and creating travel itineraries, but respondents remained concerned about security and the protection of payment details.

Phil Pomford, executive lead for Enterprise APAC at Global Payments, said consumers wanted confirmation, certification and evidence that the system was secure before delegating the final purchase decision.

Singapore recorded the strongest demand in the survey for a government-backed AI safety certification, with 34 per cent of respondents supporting the idea. It also led in preferences for a phone notification before an AI-driven purchase. In addition, 41 per cent wanted biometric approval, compared with 36 per cent globally.

The results point to practical questions for payment companies and AI start-ups. These include who would be responsible when an agent makes a mistake, how fraud would be reduced and what would happen if a user disliked an item selected and purchased by the system.

Spending limits remain modest

For now, respondents in Singapore said they would trust an AI agent with relatively small purchases. The average amounts were around S$49 for health and wellness and S$33 for a one-time digital purchase. Despite those modest figures, respondents wanted the same checks they would expect for a wealth-management transaction.

They were more comfortable with an agent carrying out a money transfer of S$400 or executing a trade worth just under S$1,200.

The cautious approach comes as Singapore maintains high levels of digital-payment use. An earlier study by PwC and the Singapore FinTech Association put digital-payment adoption at 92 per cent in 2025, alongside widespread mobile-wallet use and QR-code acceptance.

E-commerce sales in Singapore were projected to reach US$14 billion by 2027, nearly double the amount at the start of the decade. The same background noted that scam losses in the first half of 2025 reflected increasingly sophisticated schemes and business models.

Singapore's wider AI ambitions

Singapore is seeking to become a global AI hub while the industry remains dominated by companies from the United States and China. The country allows businesses to claim AI spending for tax deductions and allowances, and is committing more than S$1 billion, or US$783 million, to public AI research over the next 5 years.

An AI council chaired by Prime Minister Lawrence Wong provides strategic direction. Wong said in February that AI “must serve our national interests and our people”.

The survey suggests that Singapore's established digital-payment infrastructure may support adoption of agentic AI, but only if users retain visibility and control over transactions.

Conclusion

Singaporeans are interested in AI shopping agents, but the study shows that broader use will depend on safeguards, certification and approval mechanisms that keep people involved in the final decision.