PR teams have moved past the question of whether to use AI and landed on a harder one: how much of it should show. Drafts, pitch angles, media lists and even crisis statements are increasingly AI-assisted, yet the brands doing this well rarely look any different from the ones that are not. The ones doing it poorly are becoming easy to spot, and that visibility is starting to carry real reputational cost.
The issue at stake is not adoption, it is calibration. Communicators who lean too hard on generative tools risk producing content that reads as generic, over-polished or subtly off-voice, the exact opposite of what a brand’s communications are supposed to achieve. Getting this balance wrong is no longer a hypothetical risk. It is showing up in bylines that sound nothing like their supposed authors and pitches that read like they were sent to everyone at once, because they were.
Where AI Quietly Erodes Brand Voice
The most common failure mode is not obvious AI-generated nonsense, it is content that is technically correct but stripped of the specificity that made a brand’s voice distinctive in the first place. A CEO byline that once had a particular cadence, a few recurring phrases, a willingness to take a position, starts reading like every other CEO byline once AI smooths it into “professional” prose.
This matters more for PR than for most other content functions, because brand voice in communications is often the entire point of the exercise. A press release exists to convey a specific position in a specific tone, and if that tone becomes interchangeable with a competitor’s, the release has arguably failed regardless of how efficiently it was produced. Recent 2026 State of Journalism research found that the vast majority of journalists are now using AI tools themselves, which makes generic, AI-flattened pitches and bylines even easier for them to spot.
The Pitch Problem: Personalization at Scale
AI tools have made it trivial to generate dozens of “personalized” pitches in the time it once took to write three. The trouble is that journalists can tell, and the tells are becoming well known: pitches that reference a reporter’s beat in generic terms, that use phrasing patterns recognizable across an entire inbox, or that get basic facts about recent coverage slightly wrong. According to the State of Journalism 2026 report, the large majority of journalists delete pitches outright when they miss the reporter’s actual beat, regardless of how polished the writing is.
A few practices help keep AI-assisted pitching from reading as mass-produced:
- Use AI to draft the pitch structure, then personalize the opening line and specific hook manually, based on something the reporter actually covered recently
- Avoid running the same AI-generated pitch through minimal find-and-replace for different outlets
- Keep a running list of phrases the tool defaults to, and edit them out consistently, since journalists notice repeated verbal tics across pitches from the same firm
Crisis Communications Is Where the Stakes Are Highest
Using AI to draft a first pass of a crisis statement is increasingly common and largely uncontroversial. The risk comes when that first pass is treated as close to final, because crisis language requires a level of precision, legal awareness and emotional calibration that generative tools are not reliably equipped to provide on their own.
A statement that sounds slightly generic in a product announcement is a minor issue. A statement that sounds slightly generic, or slightly tone-deaf, during an actual crisis becomes the story itself. Human review at this stage is not optional, and any PR team treating it as optional is taking on risk disproportionate to the time saved.
Building an AI Workflow That Protects Voice
The firms managing this well tend to treat AI as a drafting and research accelerant rather than a finishing tool. AI handles the first structural pass, background research, and initial variations, while a human retains final control over tone, specific phrasing, and anything that will carry a named executive’s voice.
This distinction, between AI-assisted and AI-finished, is becoming a meaningful differentiator between firms that maintain distinct client voices and firms whose output is starting to blend together across accounts. Clients are beginning to notice the difference, even when they cannot articulate exactly what is causing it.
What This Means for the PR Industry
The industry’s relationship with AI is settling into a more mature phase, past the early debate over whether tools should be used at all and into the harder work of figuring out where the line sits for a given brand, client or moment. That line will not be the same for a product launch as it is for a crisis statement, and firms that apply a single AI policy across every type of communication are likely to get it wrong somewhere. Agility PR Solutions’ own coverage of PR technology trends tracks how these tools are reshaping workflows across the industry more broadly.
The competitive advantage going forward will not belong to the firms using AI the most, but to the ones that have figured out precisely where AI stops and a human voice needs to take over, and that have built workflows disciplined enough to enforce that line consistently.
Agility PR Solutions is a recognized leader in enterprise-ready AI-optimized PR tools and resources, including AI-powered media monitoring, media database, and media intelligence and measurement.
