ShareLinkedInXEmail
GOVERNANCE & RISKDeveloping
Opp 7Threat 8ActImmediatehigh confidence

Agency MSAs Are Dangerously Behind as AI Rewrites the Economics of Delivery

·3 min read·1 source
1

The Development

Agencies are operating a de facto SaaS delivery model — AI-automated workflows, proprietary generative tools, restructured staffing — while their Master Service Agreements remain largely unchanged. Four agency executives surveyed by Digiday in September 2026 confirmed the gap: AI has materially altered how work is produced and what it costs, but none say they are ready to rewrite their MSAs to reflect that reality. The mismatch matters because MSAs govern intellectual property ownership, indemnification, staffing minimums, output standards, and cost-pass-through rules — every clause that determines who is exposed when an AI-generated asset creates a brand safety incident, a copyright dispute, or a billing anomaly.

2

Our Take

This is not a procurement story. It is a liability story dressed as a procurement story. When an AI tool hallucinates a claim in a campaign asset, or generates imagery that infringes a third-party copyright, the MSA determines who pays. Agencies have every incentive to delay renegotiation — the current framework lets them capture AI efficiency gains without surfacing them contractually. Client-side marketing leaders who treat MSA review as an annual legal formality are, in effect, subsidising their agency's margin expansion while absorbing undefined risk. The window to reset these terms on favourable terms is now, before agencies standardise AI addenda that protect their own position.

3

What Changed

Agencies now have the operational capacity to replace significant portions of human labour with AI tooling while billing under contracts written for human-labour delivery. That asymmetry — new economics, old legal framework — creates an unpriced risk sitting inside every active agency relationship.

4

Marketing Impact

Procurement and marketing operations teams carry the most direct exposure. Existing MSAs almost certainly lack clauses covering AI-generated IP ownership, model training data disclosure, output indemnification, and subprocessor transparency — gaps that become litigation triggers the moment anything goes wrong.

5

Competitive Implication

Agencies holding outdated MSAs maintain margin protection while clients absorb unquantified IP and liability risk. Clients that renegotiate now can extract AI cost-pass-through transparency, output audit rights, and IP assignment clarity before agency-standard AI addenda harden the terms in the other direction.

6

Strategic Outlook

Expect agencies to begin rolling out standardised AI addenda in Q4 2026 as legal teams catch up with delivery reality. Those addenda will be written to protect agency IP and limit indemnification exposure. Clients who have not already initiated renegotiation will be responding to agency-drafted terms rather than setting their own.

7

The Exploit

Action Item

CMOs with active agency MSAs should instruct procurement to open a contract review before Q4 2026, specifically demanding disclosure of which AI tools are in the delivery stack, who owns outputs, and how efficiency gains are passed through on cost.

8

Source