Overview
- One major ecommerce site won a preliminary injunction that stops third‑party AI agents from using password‑protected accounts without the operator’s approval, even when a user grants the agent permission.
- NIST’s Center for AI Standards and Innovation launched an AI Agent Standards Initiative to build interoperable and secure systems for autonomous shopping tools.
- Retailers are updating terms to put consumers on the hook for AI mistakes, with Walmart’s policy for its Sparky assistant cited as a key example that underscores growing liability concerns.
- Early performance remains shaky, with industry testing around a 70% success rate and errors such as out‑of‑stock purchases and bad shipping choices risking higher return costs in a market where returns already total about $890 billion a year.
- Brands are gearing up for the shift—Mondelez is hiring a global lead for agentic commerce and some marketers are feeding verified product data to models via a model context protocol so agents can check answers against real catalogs.