Welcome to Product Cocktail, where the takes are as polarizing as a shot of Fernet—but the insights come together like a perfectly crafted daiquiri.
The Shake

May 18, 2026:
While mindlessly tapping through Instagram stories, I landed on one posted by a friend who's trying to break into the LA comedy scene. I stopped in my tracks, my eyes subject to a garish, visual assault on my senses.
In large block letters against an explosion of party paraphernalia "ALL CHATGPT FLYERS LOOK THE SAME! Especially with this font!" (You know the font, now at least.)

ChatGPT just launched a drone strike on its own position with this image gen. (Source: Leah Debene, Facebook)
I chuckled at the obvious AI slop, blissfully unaware of this phenomenon, as a new dad across the country, not tapped into the LA comedy scene (or... any comedy scene, for that matter).
Two weeks later, she reposted another version of the same flyer parody.
Then, the infamous "ChatGPT flyer" was everywhere.
LinkedIn. Local charity golf tournaments. Even at a local pizza shop on the Connecticut shore just trying to make it past the first year hump.
"We are living in a 'ChatGPT Flyer Pandemic'"
On July 10th—days after I saw the pizza shop flyer—an article hit my inbox from 404 Media claiming that "we are living in a 'ChatGPT Flyer Pandemic'" inspired by a viral Threads post from early June.
404 Media's Jason Koebler describes the visual grammar:
The art of the format is basically big, flashy bright text on dark background and an AI-generated or AI-altered image. There is almost universally a little box of generic icons in a bulleted list vaguely tied to whatever event or business it’s advertising, lines coming off of the text to emphasize whatever it’s saying, and either bolded words or underlined text and tons of arrows and checkmarks haphazardly strewn throughout.
Koebler documents an extensive infection radius: a Venice Beach sidewalk sign advertising surfing, invites to parties from friends, Fourth of July parties hosted by bars, drug deliveries in Berlin, junk hauling in South Carolina, fundraisers in Texas.
The ChatGPT flyer looked at COVID-19 and said "hold my beer."

My eyes are bleeding. (Source: Zakkai Rayne Morningstar)
The backlash
In May, I attempted to reckon with the shocking level and speed at which negative sentiment had taken hold toward consumer AI.
The same week I published that newsletter, Gen Z lifestyle content creator Gabe Escobar (idfk either, but he has 3.4M TikTok and 560K Instagram followers) posted a visceral takedown of AI on the 'gram.
"I think if you need AI to do anything, especially creative ventures, you should not be doing that thing... Because if you're using AI, you're not doing it. A computer is doing it, and you're just there... I don't know how to fly a plane, but I want to sit in the captain's seat, so I'll just use autopilot."
That video did 2M views, 203.7K likes, and 23.6K reposts.
The negative AI sentiment had been building, and in June, local businesses and events started catching strays.
"Hey if this is your flyer, I'm not going, I'm not donating, I'm not sharing. Don't ask me 🤙"
Shares outrunning reactions 3 to 1. A random Facebook post being blasted out, likely as cautionary tale to others.

Read the left column: you think it looks good because you're inexperienced. The parody got to my argument about eight weeks before I did. (Source: Jill Oliver, Facebook)
Oliver's words posted above the meme serving as a definitive takedown of ChatGPT's house style: "YOUR FLYER LOOKS LIKE GARBAGE."
In mid-July, Lyndsey Lozano put a sign in the window of Grind and Unwind, her two-month old San Francisco coffee and pastry shop: AI-generated images of drinks and food, a temporary promo to pull in tourists.
First came the online backlash (predictable). "If that's what your food looks like, I don't want to eat there." Then, a week later, someone spray-painted a single word across the front of her building.
Seriously.
Lozano took the sign down.
AI companies are rubber, local businesses are glue...
The latest backlash has manifested as anti-AI graffiti on at least 10 billboards in… San Francisco? The same place that is host to "MCP Nights" and has historically gotten away with "STOP HIRING HUMANS" bus station ads (makes you nostalgic for the days of nerds being shoved into lockers).
Mintlify's agentic bullshit billboard cleverly marked with "At least the old billionaires were philantropic".

Maybe Amodei just needs a guy name Phil on his board. (Source: SF Standard)
The AI elite buying the billboards are unfazed.
In response to one of Pylon's billboards in a $350k campaign being tagged with "fuck your AI billboard," 29-year old CEO and founder, Marty Kausas, laughed it off. "Honestly, fair."
Weeks earlier, across town, someone with another can of spray paint wrote one word on Lyndsey Lozano's pastry shop. Seriously. She repainted the building.
Kausas thought his was funny.
Jaspar Carmichael-Jack (what a name), CEO of Artisan (the Stop Hiring Humans people) thought that the "backlash is earned."
It's below the line for them.
AI backlash is projected to be a key risk factor in Anthropic's IPO prospectus. DGAF.
A Gallup survey from March noted that 71% of polled oppose a data center in their area. DGAF.
A local business buys into the promises of these executives, posting an AI-generated flyer. Unfollowed. Unsubscribed. Customer lost.
Wild times we live in
To be clear: I'm not trying to absolve every mom and pop restaurant, comedian, or bar owner who's caught shrapnel by (unintentionally) wading into the AI culture wars. I don't expect every SMB operator to be read in on consumer AI sentiment (like and subscribe, people!).
There's a difference between not knowing and willfully disregarding.
Take Wildwood Spirits, a small Seattle distillery that describes itself as "small-batch handcrafted" (we'll come back to that shortly). They've been posting AI slop to Instagram and Facebook for a while, interspersed with honest camera phone clips and genuinely well-produced real videos, which is the sad part.
In December 2025, they posted an AI-generated video of local celebrity squirrel "Sprinkles" rampaging through Seattle, like, well, Rampage. A customer questioned the use of AI. The brand's response:
"sorry that you feel that way, we are using AI in an unrealistic way because realism with AI feels deceptive. We have tried realism but it just doesn't land or feel right. We hope to see you back but also respect you decision. Have a good day."
The customer unfollowed. Asked and answered.

I’ll take a Gim and Tonic, please. (Source: @wildwoodspirits)
This doesn't seem to be a shop with no other options. Wildwood makes six spirits, distributes to Safeway, Whole Foods, and Total Wine, and ships to 35 states. They already employ people who can shoot, based on the well-produced content on their socials.
For what it's worth, the AI hasn't exactly produced breakout viral moments. Their best-performing video of the last two months is an unglamorous 7-second clip of one of their production assistants sucking whiskey out of a barrel with a glass tube: 58,100 views.
Their AI take on Jimothy the raccoon—the biggest local story of the year, in their backyard—did 1,706. The curious, slightly weird moment shot on an iPhone did 34x the views in the same month. (AI can obviously go viral, a partner cross-post with a heavy AI 'murica intro did 49k. I'm not deriving a social media strategy from two posts.)

That ain’t Jimothy, and that ain’t Ballard. (Source: @wildwoodspirits)
The reach is only one piece. Remember "handcrafted"? Craft is the entire product claim. In May 2026, part of a video they posted contained an AI-rendered shot of their gin with scrambled letters where the volume and ABV should be, and a product name reading "Kur Gim." Kur is their gin. The tool could not spell the name of the thing they make. The product page on their website isn't much better: "750rol / 49% alc. by volums."
An average-image-generator misspelled the bottle. That's not a marketing miss, it's a tool contradicting the craft they put into creating the product.
Frictionmaxxing
And in the other corner... "In the Age of AI, the Analog Restaurant Feels Radical" a mid-July romanticizing of handmade touches by Bettina Makalintal at Eater.
It's a love letter written to places like new Manhattan restaurant, Somssi, which announced itself with a postcard-sized piece of lace with the restaurant's name, opening date, and address beautifully sewn into it in script.

Somssi is a Korean word that means “a skill or a knack that you polish over time, and has a lot to do with things that you craft with your hands.” On brand. (Source: Eater)
Or Fearless Coffee in Johnson City, Tennessee (goddammit, now I'm humming "Wagon Wheel") that announced its new menu with a series of 90s-coded images that... admittedly hit me right in the feels. (It became their most-liked post ever.)
Evidence of "frictionmaxxing" a term coined by Kathryn Jezer-Morton, for rejecting technologies of convenience in favor of slower, analog processes.
Cut to: Product Cocktail, issue 5, "The AI Vibes Are Off"
I think for someone to purchase vinyl at a show, these people are walking away as fans with a piece of art. Yet it's more expensive. It doesn't sound better. It doesn't make things easier. It's the opposite of that.
I described this sentiment at the time suggesting that friction is the product. Two months later, the term for that phenomenon "escaped containment" and restaurants run it as strategy.
I cited brands like Apple releasing BTS videos of their ads to provide they didn't use AI. For restaurants, that's become a sharpie.
Somssi commissioned lace. Fearless Coffee hired a photographer. Those are choices made by businesses and called positioning.
Lyndsey Lozano put real pastries in her window because someone spray-painted her building.
They arrived in the same place, but she didn't pick it. At the time, I called the humanity premium a consumer preference—and it still is one, if you can afford to make it. Lozano got a bill.
Amodei-ity
What does the AI industry's favorite cheerleader God-King naysayer have to say about all of this AI backlash?
"I think it is fundamentally a crisis of trust. I think that ordinary people don't trust companies, governments, or the tech industry and always suspect that we are cooking up some new way to screw them over. The causes of this go back decades, and AI is just the latest iteration of it."
His fix: "actually curing cancer." Or at least, delivering on the promises to benefit the world.
For whatever it's worth, per CNBC Generation Lab, over 75% of 18-34 year olds don't trust Dario Amodei to act responsibly, and ~70% say the same of Zuck and Altman. Among the Big AI Dogs*, the one diagnosing the crisis of trust polls the worst.
He's partially right.
Regular people don't trust companies. The reasons go back decades. AI is the latest iteration of it?
That's a very convenient excuse.
Social media took over a decade to turn—Facebook launched to the general public in 2006 and the chickens didn't come home until Cambridge Analytica in 2018. AI got booed at a commencement inside of four years. Whatever this is, it isn't the slow accumulation of general corporate distrust that AI happened to inherit. It's faster and louder than its supposed predecessor.
This makes the small business problem harder to explain away. Diffuse anti-corporate grumbling would explain indiscriminate targets. This is different. People know precisely which industry they're angry at. They just can't reach it.
Nobody who unfollowed Wildwood was settling up with Anthropic. Jill Oliver didn't skip that fundraiser to send a message to OpenAI. A stranger spray-painted Lyndsey Lozano's building over a grievance with an industry she doesn't work in.
The backlash has no working path to the companies that caused it, and a perfectly functioning one to the sandwich shop.
Cure cancer, and maybe we're square. The pizza shop's invoice comes Thursday.
And Zuck, who is only a big dog in the AI spend category so far.
Who gets the bill?
Some of these small businesses have earned one, the question is how to tell from the outside.
Luke Arens—the music producer from issue 5, and interviewee most likely to be an anti-AI hardliner—drew his line somewhere specific and fair. Another artist using Suno to produce a song for sync licensing (typically ads or tv/film) was theft. A 14-year old using the same tool to make his friend a birthday song was "almost like an introduction to creativity." The line wasn't about AI, it was about whether it took someone's job.
Wildwood is the sync license equivalent. They have the capability (and quiche) to produce something real, and instead fired the slop cannons.
By that test, the pizza shop flyer is the birthday song. They probably didn't have a designer in the budget. It displaced a flyer made in Canva or Word by a stretched-thin owner.
Jill Oliver's post was shared with 3500 people. I'm not going, I'm not donating, I'm not sharing. Fair enough. That's a meaningful cost to a business thinking about how to make payroll for their staff next week. It's not going to mean shit to OpenAI, but it might cost a pizza shop a Friday.

The AI scarlet letter was delivered by the top AI house
Too much: To comply with the EU AI Act, Anthropic delivered an imperceptible (to the naked eye) watermark on output from Fable last month. Opus will carry it by the time you recover from your two day LDW hangover. Fully generated work should carry a mark—the ChatGPT-written student essay, the Ghibli-inspired AI image, the Suno song.
The problem is that Anthropic picked the cheapest way to comply with this law, apparently firing this mark on any text that comes out of their models, even a spelling correction. Last time I checked, we didn't need to narc on people using spellcheck, unless it was the Spelling Bee. The mark reports contact and reads as authorship.
Not enough: AI bystander (if there is such a thing) awareness that the mark nobody announced quietly shipped years ago. It's avoidable via precision and nuance. It doesn't measure effort or care — it measures whether you knew the tell. This gets read as an audience that didn't try, but often it means didn't know. "You're right, and that changes the beat rather than just softening it." 😉
What's the fix? I recommended how to build in a hostile marketplace: consider the counterfactual, allow opt-in, provide visible inputs, and offer corrective paths. I didn't consider proportionality: does the signal your product emits match how much of the work the model actually did?
If you ship provenance, it must express degree. A flag with one setting isn't transparency. In this economy, it's an indictment. If you ship a strong default aesthetic, you've sorted users by sophistication and may send the wrong message to their audience. Don't send a signal that puts a scarlet letter on your users.
Tip Your Bartender
Send me questions, feedback, and cocktail recipes:
[email protected].
Icons made by Icongeek26 from www.flaticon.com.


