Publishers Push Back on Agentic AI Ad Buying at Cannes

Publishers demand clarity on agentic AI economics

Publishers arrived at Cannes Lions 2025 with a pointed agenda: separate genuine innovation in AI-driven media buying from rebranded intermediaries that could extract fresh fees from an already opaque supply chain.

While holding companies like WPP unveiled video-buying agents ahead of the festival, senior publishing executives told Digiday they're focused on technical specifics—which agencies have actual builds, budgets, and timelines versus conceptual pitches. One commercial lead at a major news organization said the priority is understanding "how they would like to see our data being passed to them, and are they ready to trade in that way."

The central concern: whether agentic systems will genuinely reduce friction between buyers and sellers, or simply recreate the programmatic middle layer that publishers estimate captures up to 50% of advertising spend. "If we're not careful, every single ad-tech company is going to become an agentic ad-tech company," one global publisher executive said, warning of new per-impression AI charges—token fees—that compound across billions of ad calls.

Data control and walled gardens

Beyond transaction mechanics, publishers are scrutinizing how holding groups are consolidating data into centralized "spines." Publicis's acquisition of LiveRamp has become a focal point. One executive described concerns that what was positioned as "this neutral link across the industry" could transform into "the facilitator of a walled garden network within the Publicis environment."

The question publishers are pressing in Cannes meetings: will their inventory still surface when an agent or planner queries these systems for specific audiences? The fear is a repeat of programmatic's early days, when infrastructure was "done to publishers" without their input on standards or economics.

Yahoo COO Matt Sanchez struck a more optimistic note, saying "agentic AI for media planning and buying has shifted from concept to reality," pointing to emerging standards like AdCP and MCP designed to enable software agents to interact with ad systems without manual dashboard navigation.

AI licensing and content economics

Running parallel to agentic discussions is what Digiday described as a "quietly urgent agenda" around AI licensing and scraping. Publishers are meeting with cloud providers and platforms to assess whether charging AI systems for content access can become a meaningful revenue stream.

Amazon's recent announcement allowing publishers to charge for scraped content signals movement, but publishers remain frustrated by the scale of venture-backed scraping operations. "Everyone else is building profitable businesses on the back of our IP," one anonymous publisher said. "You can have a Napster for a bit, but we need to have a Spotify."

Publishers are using services like Azure Web IP and AWS CloudFront as proof points that paying for quality intellectual property is both technically feasible and commercially viable.

Why it matters

The agentic AI conversation at Cannes represents a critical juncture for the open web's economic model. If publishers fail to shape the technical standards and data flows of automated media buying, they risk repeating programmatic advertising's history—where intermediaries captured the majority of value while content creators absorbed the cost. With AI agents poised to become primary discovery and transaction layers, the infrastructure decisions made in 2025 will determine whether publishers retain meaningful control over monetization or become further marginalized in their own supply chain.

The New York Times Company's David Rubin captured the broader uncertainty: "In an industry that is usually marching towards what's next, many brands are searching for what to do. The one thing leaders seem to agree on is that in a world of AI, trust is likely to matter more than ever."

These details were first reported by Digiday in their Media Briefing from Cannes Lions.