During the Sept. 26 episode of Jeff Wilser’s podcast “AI Curious,” we found ourselves in a meta moment. Jeff was interviewing me about Bospar’s new GEO tool, Audit*E, when our conversation turned to the telltale signs of AI-generated PR pitches.
Jeff explained exactly how these pitches work: “The AI pitches don’t just begin with a compliment. They begin with a hyper-specific compliment. I often get: ‘Jeff, I so enjoyed your conversation with Curtis Sparrer of Bospar. I love how the two of you dug into the nuance of GEO and SEO.’ And then it goes one level deeper: ‘In a similar vein, I think you would love having Joe Schmatelli on your podcast to discuss his industries.'”
(For the record, Joe Schmatelli is not a real person. Jeff made him up to illustrate the absurdity of these formulaic pitches.)
He continued: “What’s so interesting is that as someone who’s generally bullish on AI, I think in theory, the idea of using AI to help better do research and better customize a message feels like fair game. The problem is when it’s all done in the same cookie-cutter way. Once you’re busted, once I know you’re just scouring transcripts and asking AI for a 200-word pitch, the jig is up. I’m on to you, and now I have no interest in Joe Schmatelli.”
Then Jeff predicted what would happen next: “We know that AI will be reading, watching, processing what you just said. So, we might manifest this.”
He was right.
Just 14 days later, on Oct. 10, Jeff received a pitch that began: “Hey Jeff, Your latest episode on GEO versus classic SEO was a much-needed deep dive into how AI engines shape brand visibility and discovery. The way you and Curtis examined practical GEO hygiene, especially around fixing inaccuracies and feeding reputable signals to generative models, struck a chord…”
It was the exact formula Jeff had described: a hyper-specific compliment referencing our conversation, followed by “struck a chord,” and then the pitch. When Jeff forwarded it to me, his subject line said it all: “Told you this would happen!!!”
I’m seeing it happen, too.
While I spend most of my time as a PR professional pitching journalists, I’m also on the email alias for Bospar’s podcast “Politely Pushy with Eric Chemi,” which hit its 100th episode this year. Before launching Bospar, I was a TV producer and reviewed thousands of pitches. I’ve now seen the pitching process from every angle, including this recent AI invasion.
The Ultimate Irony: AI Is Coming for PR Jobs… So, Let’s Use It Badly?
Before we dive into the pitfalls, let’s acknowledge the elephant in the room.
A recent Microsoft Research study analyzed 200,000 real-world AI conversations to determine which professions face the greatest AI impact. The results? Media and communication workers rank No. 1 among all occupational groups most likely to be impacted by AI, affecting over 602,000 professionals. Public relations specialists ranked No. 23 in individual occupation rankings.
The study’s findings were stark: “PR professionals sit at the epicenter of AI transformation, working in the most AI-impacted occupational category while practicing skills that AI has proven exceptionally capable of performing.”
So, how are PR professionals responding to this wake-up call?
Many, it seems, are doing so by hastening their own demise.
The irony is worthy of Alanis Morissette’s song “Ironic.”
Rather than using this as motivation to elevate our strategic value and human touch, some practitioners are automating the very tasks that require the most human judgment: understanding a journalist’s beat, crafting relevant pitches and building authentic relationships.
This is a self-fulfilling prophecy. If public relations professionals use AI to churn out generic, easily recognizable pitches, we’re proving the Microsoft study right.
But here’s the thing: It doesn’t have to be this way.
Here are the three rules I’m advocating we adopt.
1. Use AI to research, but write it yourself.
Every journalist is now getting pitched by AI. They know because AI pitches all sound the same.
The biggest giveaway — em dashes.
You know, the dash that’s the length of a capital M? Once you start looking for them, you’ll see them everywhere in AI-generated content.
I got so curious about this pattern that I asked ChatGPT why it uses so many em dashes. The response was revealing: “I tend to use em dashes for a few reasons: Rhythm and emphasis. Em dashes let me create a conversational pause — stronger than a comma, lighter than a period — that mirrors how people speak… Clarity without clutter. Instead of parentheses (which can make a sentence feel academic) or colons (which can feel formal), em dashes keep the reader’s attention flowing… Modern style. Many contemporary outlets — from The Atlantic to Fast Company — use em dashes liberally.”
Notice something? ChatGPT used em dashes three times while explaining why it uses em dashes. It can’t help itself. Once you know this pattern, you’ll spot AI-written pitches everywhere.
Beyond em dashes, there are other signs: an overabundance of formatting, bolding, italics and additional summaries where none are needed.
Then there are the writing crutches, cliches and formulas. My favorites: “struck a chord” and “it’s not about X, it’s about Y.”
The pitch Jeff received hit every classic marker: “struck a chord,” hyper-specific references and the carefully constructed arc from compliment to pitch.
Just this week, I received a pitch about a celebrity partnership from one of the most famous names in butter (yes, butter). When I pointed out that a significant part was written by ChatGPT, the publicist insisted it was “just a thorough pitch.” So, I ran it through ZeroGPT, an AI detection tool. The result? 35.8% AI-generated content.
Here’s the tell: The pitch described how the partnership brought together “two worlds that don’t often intersect, the farm fields of rural America and the concert halls of its biggest cities, showcasing the pride, grit and joy of rural life in a way that feels modern, compelling and deeply human.”
That’s textbook AI prose: lofty, generic and trying too hard to sound profound about… butter.
The new rule: AI can help you research a journalist’s beat and recent coverage. But the actual pitch needs to be written by a human who understands nuance and can craft a message that doesn’t sound like it came from a template. When caught using AI, denying it only makes it worse.
2. Start with substance.
You may be familiar with the well-known saying, “Flattery will get you everywhere.”
Not anymore.
In the AI era, compliments have become so associated with generic, bot-written pitches that they’re now a red flag.
Over the span of just a few days in late October, we received multiple pitches with subject lines like “Guest Appearance” and “Great guest” that all began the same way: “Hello Politely Pushy with Eric Chemi, Big fan of your podcast.”
When you’re receiving dozens of pitches with identical subject lines like “Guest Appearance,” your pitch doesn’t stand out. Each pitch then listed three generic talking points, followed by either “If this sounds like a fit, let’s lock a recording slot” or “Thought this might line up with your format.”
The pitches came from different names at what appeared to be the same operation. When we didn’t respond to one, the follow up arrived within 24 hours: “Just following up to see if you had a chance to review my previous email.”
The problems: generic subject lines that don’t stand out, generic praise that signals AI generation, zero context about why these guests matter now and contentless follow ups.
The new rule: Flattery needs to take the same path as the em dash: Don’t do it! Instead, use the subject line to convey to the recipient exactly why they should care. Lead with the news, the insight or the unique angle. Save the pleasantries for after you’ve established real value.
3. Follow up with something new.
Remember those pitches to “Politely Pushy with Eric Chemi”? The follow up arrived just 24 hours later with: “Just following up to see if you had a chance to review my previous email. I’d love to connect you with them if you think it’s a fit.”
No new information. No correction of the approach. No additional context. Just a nagging reminder of a pitch that was flawed from the start.
The new rule: Every follow up should add new information. A new data point. A connection to breaking news. A different angle. This is where AI can help: Use it to scan for recent news or trends related to your pitch, identify a fresh angle or find a timely hook. But then write the follow up yourself in your own voice. If you have nothing new to say, don’t follow up.
The Bottom Line: Be Human
Jeff predicted on “AI Curious” that AI would process our conversation and use it to pitch us. He was right. Just 14 days later, the prediction came true. But here’s what those pitchers missed: We didn’t say AI pitches were impressive. We said they were recognizable. And once a pattern is recognized, it becomes ineffective.
Even worse? Getting caught and denying it. When I showed the publicist the ZeroGPT results, the damage was already done.
Journalists have adapted faster than many PR practitioners realize. They’ve developed pattern recognition for AI-generated content. They can spot the formulaic structures, the telltale phrases, the robotic enthusiasm.
The solution isn’t to avoid AI entirely. It’s to use it as a research tool while keeping the human touch in your actual outreach. Craft pitches that sound like they came from a person (you!) who genuinely understands the journalist’s work, has something newsworthy to share and respects their time.
Showing a sense of humor doesn’t hurt either.
The Microsoft study warned us that AI is coming for PR jobs. Bad AI pitching might accelerate that very outcome. If public relations experts demonstrate that our work can be easily automated, we’re making the case for our own obsolescence.
But if public relations teams use AI strategically while doubling down on the human facets of PR, we prove our irreplaceable value.
In the AI era, human authenticity isn’t just nice to have.
It’s the only thing that cuts through the noise.
