AI Has Entered Every Layer of PR Operations
In Brief: AI tools have entered every layer of PR operations. Pitch generation, media list building, sentiment analysis, transcript summarization, brief drafting, even crisis statement first-drafts. The question is not whether agencies are using AI internally — every agency is. The question is which uses are operationally and ethically defensible and which are quietly destroying brand quality. After 25 years running one of the largest privately-owned communications firms in the world, here's where I draw the line.
Key Facts · As of May 2026
| Use Case |
Current Discipline |
| Media list intelligence | Acceptable with human verification |
| Pitch drafting | Acceptable as starting point; never as final |
| Sentiment analysis | Acceptable directional input |
| Brief drafting | Acceptable with senior edit |
| Crisis statement first drafts | Risky — never deploy without senior human edit |
| Client deliverable generation | Disclose to client; never represent as fully human |
| AI-generated content placed as earned media | Unacceptable — corrodes the citation graph |
How PR Agencies Are Actually Using AI in 2026
Every agency is using AI. The ones telling you they aren't are lying — or behind. The question is how. The industry data confirms it — adoption of generative AI across PR is now near-universal; the gap between leading teams and the rest is the operating model around it, not the tool count. See Where Big PR Actually Uses AI for the numbers. The full workflow — six stages, each with an AI tool that collapses the slow part, each with a person who owns the decision — is documented in The AI PR Stack.
The defensible uses cluster in three categories.
First, intelligence gathering: media list construction, beat-mapping, reporter sentiment analysis, message-track listening. AI is good at this because the work is high-volume pattern recognition. Human analysts still verify outputs, but the time savings are real.
Second, drafting and summarization: meeting notes, transcript compression, first-pass briefs, internal status reports. AI handles the connective tissue of agency operations efficiently. Quality holds because senior staff review and edit before anything reaches the client.
Third, optimization: A/B testing pitch subject lines, refining message frameworks, surfacing patterns in coverage analysis. AI is faster than human analysts at this and produces directionally useful output.
The non-defensible uses also cluster.
AI-generated content placed as earned media without disclosure. AI-generated executive statements deployed without human authorship verification. AI-generated client deliverables represented as the product of senior human strategy. Each is a brand and citation-graph contamination event. The disclosure architecture that keeps journalist trust intact is documented in AI-generated content disclosure, and the underlying failure modes across the industry are laid out in The Perils of Over-Reliance on AI in PR.
Where AI Use Crosses Into Brand Risk
Three thresholds define the risk.
The Disclosure Threshold
If the client doesn't know AI was used in producing the deliverable, that's a problem. Agencies should be disclosing AI use in scope-of-work documents, in deliverable headers where appropriate, and in master service agreements.
Brands paying for "senior strategy work" deserve to know what fraction was AI-generated.
The Authorship Threshold
If an executive byline or statement was AI-drafted and the executive didn't review with substantive edits, the brand is exposed.
The "I didn't write this" defense when a hallucinated claim surfaces is not a defense; it's evidence of inadequate review.
The Citation Threshold
This is the one most agencies still don't see.
AI-generated content placed at scale across owned channels degrades the brand's citation surface inside AI engines. Engines detect synthetic-feel patterns. Brand authority signals get reduced. The short-term content velocity gain is paid for in long-term retrieval visibility. The category-level scoreboard is the State of AI Search 2026, and the full method for playing on it is in the Generative Engine Optimization canon.
What Agency-Client AI Disclosures Should Actually Look Like
Three components matter.
Scope-of-Work Disclosure
A scope-of-work disclosure that names which work products are AI-assisted, AI-drafted, or fully human. Specificity matters.
"AI may be used for research and draft generation, with senior review and revision" is operational language.
"We use AI" is not.
Deliverable-Level Disclosure
Reports, briefs, and analyses that incorporate AI outputs benefit from a footnote naming the use. Clients increasingly want to know, and regulators are increasingly asking. The tooling questions that determine whether the underlying data terms even support this discipline are covered in AI Procurement for Communications Teams.
Quarterly AI Practice Reviews
Quarterly is the right cadence.
Which tools are in use. What data flows through them. What client-confidential information is going into third-party AI systems.
The agencies that can answer this clearly are the agencies winning enterprise clients in 2026. The RFP process that decides which ones win those enterprise mandates is the subject of most PR RFPs hire the wrong agency.
The AI Usage Line My Firm Draws
I'll be direct.
AI is permitted across operations. Media intelligence, draft generation, transcript work, sentiment analysis, internal documentation — yes. The productivity gain is real and not optional in a competitive agency market.
AI is not permitted to generate any content placed under a senior executive byline without that executive's substantive review and edit. The byline is the brand. Selling AI-generated thought leadership as executive insight is a credibility risk I will not take on behalf of clients I'm paid to protect.
AI is not permitted to generate content placed as earned media. The trade press relationship is the foundation of an agency's earned media function. Sending AI-drafted pitches at scale to working reporters degrades that relationship for every client we represent. The industry's underlying credibility problem — what has to be rebuilt if agencies are going to hold enterprise trust through the AI transition — is analyzed in PR's trust problem with itself.
AI is not permitted to generate client-confidential analysis on third-party platforms without contractually-verified data handling. Most consumer AI tools do not meet the standard. We use enterprise-grade infrastructure where data residency, retention, and access controls are documented.
The Read
AI use in PR operations is not optional. The question is whether it is governed.
Agencies with documented AI practice, clear client disclosures, and enforced authorship discipline are the agencies that will hold enterprise client trust through the transition.
The next crisis in agency-client relationships will not be a missed pitch. It will be an AI-related trust failure.
Build the practice infrastructure now.
The AI-in-PR cluster
