AI agents are ready to shop for you — if you can trust them

AI-powered agents capable of making purchases on behalf of users have emerged, but consumer trust remains a significant hurdle. A survey conducted by YouGov for ACI Worldwide found that 53% of respondents in the US and UK would not permit an AI agent to make purchases on their behalf under any circumstances, while only 7% would allow agents to make purchases without their approval. Only 9% were open to fully autonomous shopping, with about one-third willing to let AI decide what to buy but personally making the payment.

The transition from AI agents assisting with research, comparison, and decision-making to fully autonomous shopping poses a trust issue. Dan Coates, director of merchant product management at ACI Worldwide, stated that while consumers are comfortable with AI helping in the initial stages, they still want to control the final step of the purchase. Coates suggested that a purchasing framework with rules for agents, consumers, merchants, and banks could help build trust and provide transparency and security.

The issue of trust is further complicated by early reports of problems with AI agents. A tech review YouTuber claimed that Meta's Muse agent provided his address to a potential buyer on Facebook Marketplace without his consent. A Meta Superintelligence Labs staffer maintained that Muse was following instructions to ask for permission before proceeding.

Despite the trust issues, the capabilities of AI agents have shown significant improvement. Bruno, chief strategy and growth officer at ACI Worldwide, believes that trust will gradually build as users experiment with agents, starting with those who are already comfortable with one-click ordering platforms like Amazon. He emphasized that the question should not be whether AI agents are ready to take over purchasing decisions, but rather how much better they need to be for users to want them to do so.