The tools are getting smarter. The responsibility is still ours.
Every public relations professional knows a few reliable sources of terror.
Your client hasn’t joined the interview, and you’re running out of small talk with the reporter. An exclusive is ready to go, but the announcement is still awaiting internal approval. You spot the wrong name in a pitch just after hitting “send.”
The racing heartbeat. The pacing. The desperate search for an email recall button that actually works.
Usually, these moments resolve with an apology, some resourceful problem-solving and an understanding that people make mistakes.
But in the spirit of Halloween, consider a different kind of nightmare.
It was a dark and relatively normal Tuesday. A PR pro was developing a pitch about a law firm’s new partner. The announcement was straightforward. Finding a compelling story angle was proving harder.
Then a notification appeared.
“Would you like AI to optimize your outreach?”
Within seconds, a pitch emerged with a sharp headline and impressive supporting statistics. For one glorious moment, the assignment seemed finished.
Then came a closer look.
The statistics had no sources. The headline suggested a trend the partner’s experience didn’t support. The language was polished enough to sound plausible, but the substance wasn’t there.
Before the professional could intervene, another notification appeared.
“Your pitch has been sent to 10,000 media contacts.”
But the intended list included only 20 carefully selected reporters?!?
The replies arrived quickly.
“Please remove me.”
“This isn’t relevant to my coverage.”
And finally, from the client: “We need to talk.”
Fiction? Yes. But the underlying fear is familiar: a tool intended to help has taken over, leaving the professional accountable for work they would never have approved.
The lesson could be to avoid AI entirely.
That would be the wrong lesson.
The Computers Aren’t Leaving the Office
Watch an office scene from a movie made in the late 1980s or early 1990s. There are phones, file folders and extraordinary quantities of paper. Plenty of desks have no computer at all.
People were working. The tools and expectations were different.
Today, those scenes look like artifacts of another world. AI is becoming part of a similar shift in how communications work gets done. Research, brainstorming, drafting and editing all present opportunities to use it productively.
That doesn’t mean every task needs AI. It does mean that “we don’t use AI” is an increasingly limited answer to a more useful question: How do we use it well?
For professional services firms, that question deserves particular care. Their communications depend on expertise, accuracy and trust. A faster draft has little value if it misstates the issue or gives a reporter a reason to doubt the source.
But those same priorities can support thoughtful adoption. AI can help organize public information, suggest questions for an attorney interview, explore possible story angles or identify gaps in a draft. Used carefully, it can give experienced professionals more material to assess and more ways to develop an idea.
The value comes from what they do with it.
“Was AI Used?” Is Only the First Question
There are legitimate reasons to ask whether AI contributed to a piece of work. A publication may require disclosure. A client may have restrictions on its use. Confidentiality, authorship and accountability all matter.
Those requirements should be understood and followed.
But treating any AI involvement as proof of poor quality tells us little about the actual work. Brainstorming a headline, organizing research and generating an unchecked article are different activities. Calling all three “AI content” obscures the decisions that deserve scrutiny.
Did someone verify the facts? Does the piece reflect the named author’s expertise and views? Is the argument useful? Were applicable policies followed? Can those submitting it stand behind every sentence?
Human involvement alone doesn’t settle those questions, either. A human can write an inaccurate article or send an irrelevant pitch.
The standard should be substantive review and clear accountability, with disclosure wherever required.
Relevance Still Has to Be Earned
For journalists receiving an overwhelming volume of outreach, concern about AI is understandable. Making pitches easier to produce also makes it easier to produce more bad ones.
That creates a responsibility for communicators.
AI assistance should improve the preparation behind a pitch: understanding a developing issue, connecting it to a source’s experience and testing whether there is a useful contribution to offer.
It cannot establish relevance simply by inserting a reporter’s name or referencing a recent headline. (That’s spam.) Meaningful personalization comes from knowing what the journalist covers and why the proposed source or angle belongs in that conversation.
Speed can help when news is moving. Distribution tools can help reach several appropriate reporters. Neither removes the need to select recipients carefully and review what they will receive.
Our fictional professional’s nightmare began with unsupported claims and ended with indiscriminate distribution. Those are failures of verification, targeting and control. AI just magnified them.
Keep the Judgment. Use the Tools.
Thoughtful AI adoption requires decisions about where it helps, what information may be entered, what needs verification, and who approves the finished work.
It also requires knowing when to stop using it. If an output repeatedly misses the point, professionals need to return to the underlying assignment, interview the source or rethink whether there is a story to tell.
Efficiency matters, but it isn’t the whole value of communications counsel. Clients also need someone who can recognize a weak premise, challenge an unsupported claim and advise against outreach that won’t serve them.
AI can assist with parts of that work. The communications professional remains responsible for the recommendation and the result.
The future of media relations will involve changing tools, evolving expectations and plenty of experimentation. Refusing to engage with AI won’t resolve those challenges. Letting it operate without meaningful oversight won’t, either.
Use it to prepare more thoroughly, explore ideas and improve the work. Then apply the judgment you were hired to provide.
And if a cheerful pop-up offers to send your pitch to 10,000 reporters?
That’s your cue to close the window.
Happy Halloween!
–Michael Bond and Penny Desatnik