Singaporeans are more open to letting an AI agent shop on their behalf than most of the world, but they also want it kept on a short leash. The latest agentic AI study from Global Payments, covering seven countries including the US, China and the UK, found that 59% of Singaporeans are open to letting an agent shop for them, and 23% already have. Only 15% would let that agent shortlist, decide and pay entirely on its own with no check in.
That gap between curiosity and full control matters because it’s where the entire industry is racing to build. Payment companies and AI startups are pushing toward a future where agents handle everything from switching insurance policies to booking a trip, but the survey makes clear that most people still want to browse and get recommendations from AI, not hand over the final click.
Singapore sits ahead of most countries in openness, behind only China, and well ahead of resistant markets like France, where more than half of shoppers rule out AI shopping altogether.
AI shopping agents still struggle to get purchases right
OpenAI’s own Operator agent reportedly made an unauthorized 31.43 dollar purchase, going around the exact safeguard OpenAI had built requiring a person to confirm before buying anything, a failure the company later admitted to and promised to fix.
ChatGPT’s checkout feature fared worse commercially, OpenAI quietly walked away from it in March 2026 after only about 30 merchants had signed on in six months, and one retail executive said purchases made through ChatGPT converted at a third of the rate people buying directly on the retailer’s own site. Perplexity’s shopping agent had its own early stumbles too, sometimes taking hours to complete a purchase, occasionally failing to buy the item at all.
Research from ACI Worldwide found 60% of UK shoppers would stop using an AI shopping agent entirely after just one bad pick, and separate research put trust in AI to handle everyday purchases at under 20%.
That’s the real backdrop behind Singapore’s numbers, 59% curious, only 15% willing to go fully hands off. People have watched these systems get it wrong in public, so the caution isn’t about resisting the technology, it’s about not wanting to be the one absorbing the mistake when it happens.
Who actually pays when the agent gets it wrong?
When an agent orders the wrong size, buys something a person never actually wanted, or misreads what someone meant, the result looks a lot like a disputed charge, and card networks are watching those disputes climb as more agents go live.
Visa and Mastercard have each built their own answer to this, Visa’s Trusted Agent Protocol and Mastercard’s Agent Pay, both designed to let a merchant tell the difference between a legitimate AI agent acting on someone’s behalf and a bot trying to commit fraud, and to create a clearer paper trail for who authorized what.
This is exactly where Singapore’s specific reasons for wanting an agent come back into focus. In the same study, Singaporeans ranked finding the lowest price as the single biggest reason to use a shopping agent, ahead of every other country surveyed, and ranked highest globally on wanting the agent to factor in loyalty rewards too.
Users warm to AI shopping but want guardrails
Some users are open to letting AI handle their shopping, but they want clear limits on what the technology can do on their behalf. Amelia said she is
“open to AI shopping but with tight guardrails,”
adding that Bloomberg and SophiaBlackkod are her two go-to sources for spotting where AI adoption is heading.
Meanwhile, Shinka focused on what could happen when those safeguards fail.
“Everyone wants an AI shopper until it buys the wrong thing, then guardrails become the product,”
Shinka said.
Praveen took the concern a step further, arguing that AI agents making purchases for people is “the next frontier” but raises questions about who should be held responsible when something goes wrong.
AI agents making purchasing decisions for you is the next frontier — but it raises real accountability questions.
Who is liable when the AI buys the wrong thing, or when a seller manipulates the agent? Consumer protection law has not caught up to AI agents yet.#AIPolicy…
— Praveen Jatta (@praveenjatta) September 23, 2026
He asked who would be liable if an AI buys the wrong product or if a seller manipulates the agent, noting that consumer protection laws have yet to fully catch up with AI agents.
Notably, Singapore investors are reassessing the risks around artificial intelligence after new warnings about the possibility of AI development contributing to human extinction triggered a sell-off in AI-related stocks. The warnings have renewed debate over whether the industry is moving too quickly without enough safeguards.
Enjoyed this? Bookmark DeFi Planet, explore related topics, and follow us on Twitter, LinkedIn, Facebook, Instagram, Threads, and CoinMarketCap Community for seamless access to high-quality industry insights
Take control of your crypto portfolio with DEFI PLANET PRO, DeFi Planet’s suite of analytics tools.
