Notifications
Clear all
Topic starter
10/07/2026 9:23 pm
TL;DR: AI is changing how consumers discover and evaluate products, but trust still limits deeper adoption at Shoptalk 2026, according to Signifyd. 49% use AI as a research assistant, 17% of product searches start with AI, and 80% of consumers distrust AI-driven outputs. The retail funnel is compressing, not disappearing, and visibility now depends on being selected by AI as well as humans.
NHIMG editorial — based on content published by Signifyd: Shoptalk spring 2026 on AI, trust, and the future of commerce
By the numbers:
- 49% of consumers use AI as a research assistant, but only 14% use it as a decision maker.
- 17% of product searches now start with AI assistance, while 27% still rely on traditional search engines.
- 69% of purchases still happen on retailer websites, and only ~10% happen directly through AI interfaces.
Questions worth separating out
Q: How should organisations govern AI-assisted shopping journeys?
A: Organisations should define where AI can assist, where it can recommend, and where it can act on behalf of a user.
Q: Why does AI-mediated commerce create identity risk?
A: AI-mediated commerce creates identity risk because the system may influence or carry out actions without a clear human present at every step.
Q: What do security teams get wrong about AI visibility in commerce?
A: Teams often treat AI visibility as a marketing or search issue, when it is also a trust and governance problem.
Practitioner guidance
- Define policy for AI-assisted purchasing flows Map where AI can assist, recommend, or act, and require explicit consent, scope limits, and audit logging for each step in the buyer journey.
- Validate the provenance of recommendation inputs Review structured data, reviews, merchant feeds, and external signals that influence AI discovery so manipulated or impersonated sources do not shape purchase decisions.
- Separate human intent from delegated automation Apply identity and access rules that distinguish a shopper acting directly from an assistant acting on their behalf, especially for account changes and checkout actions.
What's in the full article
Signifyd's full post covers the operational detail this post intentionally leaves for the source:
- The underlying Shoptalk session references and speaker context behind the AI-assisted commerce findings.
- The full set of consumer behaviour comparisons across discovery, validation, and purchase channels.
- The commerce visibility observations that show how on-site and off-site signals affect AI selection.
- The broader fraud and returns context that sits behind the retail trust discussion.
👉 Read Signifyd's analysis of Shoptalk 2026, AI-assisted commerce, and trust →
AI-assisted commerce and trust: what it means for retail teams?
Explore further
10/07/2026 9:45 pm
AI-assisted commerce is not autonomous commerce, but it is already an identity governance problem. The article shows that consumers are using AI to research and narrow choices long before they hand over final judgment. That means the trust boundary is shifting upstream, where identity, content authenticity, and recommendation integrity all matter. Practitioners should treat AI-mediated discovery as a governed access path, not a marketing feature.
A question worth separating out:
Q: Who is accountable when an AI assistant influences a bad purchase or fraud event?
A: Accountability should remain with the organisation that set the policy, exposed the data, and allowed the delegated workflow to operate. Human oversight does not disappear just because an AI assistant shaped the journey, so governance, fraud, and customer teams need a shared control model.
👉 Read our full editorial: AI-assisted commerce is compressing the retail funnel, not replacing it
