We are the children of concrete and steel
This is the place where the truth is concealed
This is the time when the lie is revealed
Everything is possible, but nothing is real
—Living Colour
A few days ago, I thought Rhea Seehorn might have sent me a private message.
That is not the sort of sentence I expected to write, and it is certainly not the sort of sentence a sensible man should leave sitting around unattended. But there it was.
This is a story about celebrity impersonation, but it is also about the dark side of artificial intelligence: its ability to make fraud feel personal. The scammer no longer has to shout from the shadows. With the right tools, he can whisper in the voice of someone you admire.
I became a fan of Rhea Seehorn during her starring turns on Better Call Saul and Pluribus, and awhile ago I started following a Facebook “Rhea Seehorn” page. It posts first-person announcements about her acting projects, photos of her at public appearances, awards, and the usual fan-page material. Recently, it shared an announcement about an award she was receiving, and I left a comment:
Well, you are pretty amazing. I remember watching some interviews with you while Better Call Saul was running and noticing differences between your personal mannerisms and the speech patterns from Kim Wexler and realizing, “Oh, she’s acting in that role.”
And thanks for saving Bob Odenkirk. I kinda like him too.
A day later, I received a private Messenger reply from “Rhea Seehorn”:
That’s incredibly kind of you to say, thank you. It always means a lot when someone notices the little details like that, because those are the things you quietly hope land without ever being too obvious. And as for Bob believe me, I like him too. Quite a bit. I’ll happily take partial credit for keeping him around, but he makes it very easy to show up and do the work.
Really appreciate you taking the time to say this. Where are you texting me from?
It was a good answer. Maybe too good.
It hit the right notes: gracious, intelligent, actorly, just personal enough to make me feel noticed. There was no clumsy sales pitch, nothing that seemed overly suspicious, and I thought maybe it was actually her or perhaps an assistant who handles her social media.
So I answered. I said I was in Wisconsin, thanked her for the reply and added a longer note about how different Seehorn seems in interviews from Kim Wexler — more relaxed, cheerful, spontaneous — whereas Kim Wexler is tightly wound, guarded, and repressed. Her guarded quality, I said, made the emotional breaks in Kim’s composure all the more powerful: Howard’s murder, the crying scene on the airport bus, the awful moral reckoning at the center of Better Call Saul.
Then “Rhea” replied:
Thanks for such a thoughtful note. and hello to Wisconsin. I really appreciate what you said about Kim versus me in interviews. That contrast was very intentional in the writing and something I leaned into as an actor. Kim’s restraint is such a big part of her, so when emotion does break through, it’s meant to hit hard. And thank you for noticing the craft in Better Call Saul overall. That kind of attention from viewers means a lot.
You seem like an interesting person to have a conversation with, I would like to know more about you
That was when the cigarette smoke curled under the door.
The message was still subtle. It was not cartoonishly fake. The tone was plausible enough that some part of me wanted to believe it, but the ending had the shape of a hook: You seem like an interesting person. I would like to know more about you.
A working actress with an active career and a large fan base does not normally invite random men from Wisconsin into private conversation because they left a nice comment about Kim Wexler. Celebrities can be gracious to fans, but there is a professional boundary there, and for good reason. They have stalkers, obsessives, scammers, privacy concerns, and limited time.
The message didn’t smell like Hollywood. It smelled like reconnaissance. Then the plot acquired a second suspect.
Not long after my exchange with “Rhea Seehorn,” I received another Facebook message — this time from an account identifying itself as Patrick Fabian, the actor who played Howard Hamlin on Better Call Saul. This “Patrick” also invited me into a private chat.
At that point, coincidence had left the building. One possibly fake Better Call Saul cast member sliding into my direct messages was suspicious. Two was a pattern.
I reported both accounts to Facebook through its Support Center, identifying them as possible impersonators, but Facebook declined to take action. As of this writing, both accounts remain active. That may be the most discouraging part of the story: the scam did not depend on fooling Facebook’s systems. It only had to survive them.
After a little Googling, I came across a Wikipedia article on celebrity impersonation scams, an emerging category of fraud in which scammers pose as actors, musicians, athletes, entrepreneurs, or other public figures in order to extract money, personal information, or emotional commitment from victims. The impersonations have included Elon Musk, Brad Pitt, Keanu Reeves, Oprah Winfrey, Taylor Swift, Bill Gates. Some scams involve fake cryptocurrency opportunities. Some involve fake giveaways. Some involve romance. Some now use deepfaked audio or video to make the impersonation more convincing.
The Federal Trade Commission warned years ago about scammers impersonating celebrities on social media. More recent reports describe AI being used to make these scams more polished, more scalable, and harder to detect. The FBI has warned that generative AI can help criminals produce more believable messages by reducing the spelling errors, awkward phrasing, and other tells that once made fraud easier to spot.
That is exactly what made this encounter interesting. The messages I received were not wildly implausible. They were not brilliant, either. They were somewhere in the uncanny middle: smooth enough to be flattering, generic enough to be scalable, and just personalized enough to keep the conversation going.
The pivot was the key.
“Where are you texting me from?”
“I would like to know more about you.”
Those are not innocent questions in this context. They are “social engineering” probes. They move the interaction away from a public fan page and into a private channel. They invite the target to reveal location, interests, emotional state, loneliness, family situation, finances, hopes, wounds, and vanities. They build what security people call trust and what con artists call the mark.
Had I continued, I doubt the next message would have been, “Please send $500 in Apple gift cards.” The better scam is patient. It starts with attention. Then sympathy. Then intimacy. Then maybe a “private fan club.” Or a charity, an investment opportunity, a personal emergency, or a promise of a meeting that requires a fee, a donation, a verification payment, or some other small compromise that opens the door to a larger one.
The “Dear Abby” advice column recently featured a letter from a mother whose 41-year-old son believed he was in an online relationship with a famous 23-year-old actress. He was sending money every week by gift card. Abby’s advice was blunt: famous actresses do not need gift cards from admirers.
It is tempting to read a story like that and feel superior. How could anyone fall for it?
But I remember the little jolt I felt when that first message appeared from “Rhea Seehorn.” I know better than to fall for scams, but still I felt it. Someone I admired had noticed me. Someone whose work I respected had answered. The hook worked before the intellect had time to inspect the line.
That is the uncomfortable part. You do not have to be stupid to fall for this. You just have to be human.
And machines are getting very good at being human-adjacent, while fraudsters keep inventing new ways to exploit our human desire for connection.
In my day job as a site reliability engineer, I deal with security situations. I spend a fair amount of time thinking about systems, failure modes, suspicious patterns, and the difference between legitimate traffic and malicious behavior. Fortunately, my security instincts kicked in before I wandered too far down the alley.
But the encounter also reminded me how cybersecurity is changing.
We all remember the old “Nigerian prince” emails: clunky, melodramatic, riddled with typos, and easy to mock. They were a numbers game, a cheap net dragged through the ocean in hopes of catching the most vulnerable fish.
Large language models have changed the economics of the con. Scammers no longer need to send one bad message to a million people. They can hold a million tailored conversations. They can respond to what you wrote. They can flatter your taste, mirror your language, and nudge you gently toward the next disclosure.
They can scale intimacy.
In technical domains such as my job, we already see versions of this problem. The brute-force attacks of yesterday are being supplemented by “low and slow” attacks that imitate legitimate behavior. Instead of smashing against the front door, attackers test the hinges, watch the rhythms, and look for gaps in the way real users move through a system.
My little exchange with “Rhea” was the social version of the same thing. It was a conversational probe. The vulnerability being tested was not a server port or an API endpoint. It was vanity. Curiosity. Loneliness. Admiration. The ordinary human desire to be seen.
That is why the old advice — “watch for bad grammar” — is no longer enough. The grammar is getting better. The pacing is getting better. The emotional simulation is getting better.
The better warning sign is the boundary violation.
A celebrity moving you into private messages.
A stranger creating sudden intimacy.
A famous person asking where you live.
A warm conversation that somehow starts gathering personal details.
A relationship that feels flattering before it feels earned.
Those are the tells.
I do not know who was behind the “Rhea Seehorn” messages I received. It may have been a human scammer, an AI-assisted operation, or some hybrid of the two. But the pattern was familiar enough.
The synthetic starlet stepped out of the shadows, offered me a smile, and asked where I was from.
For a moment, I wanted to answer.
Then I remembered where I was.
If a celebrity invites you into their private DMs to “get to know you,” you are probably not in a Hollywood story. You are in a penetration test.
And the best way to maintain site reliability — and personal reliability — is to terminate the connection.
Leave a Reply