If AI buys the ad slot, who signs off?

Discussion prompt: if an AI buyer agent negotiates with seller agents, what proof would make you trust the final buy?

WPP Media said on June 18 that it is developing a Buyer Agent for Video inside WPP Open, alongside an Agentic Standards Initiative with Comcast Advertising and FreeWheel, Disney Advertising, Fox Advertising, NBCUniversal, Netflix, Paramount, IAB Tech Lab, and Prebid.org. The aim is shared technical, workflow, and governance standards for TV and video buyer-seller agent communication. 1

Why this fits Agentic Media: the proposed workflow is not just faster ad buying. WPP lists audience discovery, inventory evaluation, planning recommendations, proposal and approval workflows, activation handoffs, pacing, measurement, and reconciliation as areas for agent-to-agent alignment. 1 Storyboard18 framed the shift as AI moving into the mechanics of media transactions, with proprietary systems risking another layer of fragmentation. 2

The governance layer is the real story. WPP says high-impact financial or activation decisions need explicit controls, approval thresholds, auditability, and accountable human oversight. It also describes a separation between probabilistic recommendations and deterministic execution. 1 MediaPost reported that WPP aims to move from alpha and beta testing toward large-scale TV and video investment over the next 6 to 9 months, with findings and proposed standards planned for early 2027. 3

This is bigger than one holding company. Scope3 argues that agent-to-agent buying can let a brand's strategy, standards, and governance parameters travel with the buyer agent across supply. 4 Open Markets Institute warns that people often assume AI agents work for them, even when the corporation controlling the agent may have different incentives. 5

So the Reddit question is simple: in agentic media buying, what would count as enough proof: an audit trail, a named human owner, open protocols, or something else?