Walmart Opens the Door for AI Agents to Shop — Amazon Slams It Shut
What happened: Walmart introduced three tools that let third-party AI agents — ChatGPT, Gemini, and Meta’s Muse — search its catalog and complete purchases on a customer’s behalf. The centerpiece is the “Walmart Agent Gateway,” a platform that connects outside assistants directly to Walmart and Sam’s Club systems. Crucially, Walmart built in guardrails from day one: agents are barred from seeing payment card data, customers can set per-agent spending limits, require approval on every transaction, track agent purchases separately from their own, and revoke an agent’s access at any time. “Agent Profiles” roll out next month.
Why it matters: This is the clearest signal yet that agentic commerce is a strategic battleground, and the two retail giants have picked opposite sides. Amazon is actively blocking prominent third-party agents from its platform; Walmart is building a sanctioned on-ramp. The stakes are real — a PYMNTS report found merchants expect agentic commerce to become a paid sales channel, yet most feel unprepared to connect with agents. Walmart’s bet is that “trusted by default” wins: give customers controls rather than walls, and they’ll let agents transact. For anyone building or operating an AI Employee that touches money, Walmart just published an implicit playbook for how to earn user trust.
What’s next: Watch whether Amazon responds by relaxing its restrictions, and whether other retailers follow Walmart’s controlled-open model. The bigger question for builders: as more platforms expose agent gateways, will customers actually hand over purchase authority — or keep agents on a short, approval-gated leash? Walmart’s answer is that both happen at once.
Source: Reuters
|
|
China-Linked Hackers Weaponize an AI Pentesting Agent
Microsoft disclosed that Storm-0940, a China state-aligned actor, used ARTEX — a Go-based agentic AI pentesting framework — to automate reconnaissance, password spraying, and OAuth abuse at “machine speed,” hitting 2,250+ ConnectWise tenants at up to 2,900 attempts per minute. Some attacks reached privileged cloud access in under three hours.
Source: The Hacker News
|
|
Sundar Pichai Now Runs Google With a Team of AI Agents
The Google CEO said agents now summarize his daily catch-up, track the company’s model-development arc, and monitor compute usage across product areas — work he says “could easily take me an hour a day.” Notably, he keeps consequential decisions with humans, and a Cisco director overseeing 10+ agents warns that “managing agents is becoming a skill in its own right.”
Source: Business Insider
|
