chatgpt

AI Comedians Steal Instagram Jokes to Promote OnlyFans

Posted on by clementon
AI Comedians Steal Instagram Jokes to Promote OnlyFans

I was scrolling when I noticed a comedian’s punchline echoing back at me from a stranger’s account. The gesture, the pause, the setup — replayed with a different face and half a million followers. You realize, fast and cold, that the platform has been taught to want one thing and one thing only.

I’m writing from the seat of someone who follows tech and comics closely, and I want you to hear what matters: this is not just theft, it’s a new growth loop for paid pages. You’ll recognize the players — OnlyFans, Fanvue, Instagram, a16z — and the names: Justine Moore, Rosie Thomas, Olivia Carter, George Brett Olson. They sit at the center of a story that is more persuasion engineering than performance art.

On a feed: a clip that matches a comic’s set, frame for frame

I watched Tiffany Sloane — @tiffany.bbyx on Instagram — perform what looked like a working comic’s short set. The earliest post is September 11; the account drew ~375,000 followers within weeks and posted over 120 clips by October 5. The videos wear an Instagram watermark and feel like the short stand-up snippets comedians and bookers use for discovery.

What’s different is the performer. The face, body, even the voice may be AI. The bio calls the account an actress and hides a link to OnlyFans in highlights, not the main profile. OnlyFans rules require clearly labeled AI and verified creators, but accounts like this sit in a gray space — not labeled, not honest.

How are AI accounts stealing jokes on Instagram?

Here’s how the mechanics look when you pull them apart: somebody takes a recorded set from a live comic, runs it through an AI pipeline that swaps a new face or body onto the clip, posts it as original short-form content, then points viewers to a paid page on OnlyFans or Fanvue. Fanvue permits disclosed AI creators; OnlyFans expects disclosure. The net result: an AI persona becomes a marketing funnel.

That’s a marketing loop rewarded by the algorithm. You and I can see why comedians are furious: Olivia Carter told Straight Arrow reporter Rosie Thomas she kept getting copies of her personal-trainer bit, and Sarah Purdum’s Daniel Day-Lewis joke showed up on the same bot accounts. George Brett Olson posted a side-by-side and said the platform’s reporting tools didn’t fix anything.

At a journalist’s desk: a reporter matches multiple stolen sets to bot accounts

Rosie Thomas mapped several accounts and found the same pattern: follower growth, short-form clips, and a story link to subscription pages. One page’s bio read “19, and yeah, the body’s real.” Another directed to Fanvue, which explicitly allows disclosed AI. Thomas found that comics were blocked from viewing some of the stolen content — a tell that the operators know who they’re ripping off.

And then a16z’s Justine Moore posted that she’d “stumbled on a genius OnlyFans marketing hack.” Her tone was neutral, even admiring, which is itself an authority cue: a prominent investor celebrating a tactic makes it feel legitimate to startups and creators who follow her. Michel de Cryptadamus fired back on social media, arguing it’s deranged to endorse deceptive influencer strategies. The debate is public and bitter.

Is OnlyFans allowing AI content?

OnlyFans’ public policy says AI must belong to a verified creator, must clearly feature that creator, and must be identifiable as AI — #ai is the recommended flag. Fanvue’s policy is friendlier to disclosed AI creators. What we’re watching are accounts that skirt those rules by hiding subscription links in highlights and by not labeling the content as synthetic.

The scale makes this a platform problem. Tiffany Sloane’s clips pulled millions of views, per Moore’s post. When a tactic works, others copy it. The long tail of imitation is where reputations and livelihoods get eaten away.

On a product roadmap: labs demo face-and-voice models that pass for human on calls

Tavus launched Griffin, which it bills as a human interaction model that watches, listens, and generates a face and voice in the same loop. In their preview study, nearly half of participants judged a Griffin-Lite persona on a one-minute video call as real. Saagar Enjeti joked that porn companies will buy it immediately; the joke reads as a prediction.

Griffin’s demo is a preview held back for safety and disclosure work. But the direction is plain: live face-swapping and real-time synthetic personas are now feasible. Add monetization, and you have a direct pathway from algorithmic virality to subscription revenue — a funnel that can be abused.

Two facts you should hold in your head: one, the AI industry is trained on scraped material — a reality highlighted in internal Microsoft memos leaked during litigation with OpenAI where Brent Hecht called scraping “an astonishing theft of unprecedented proportions.” Two, platform incentives reward attention, not authenticity.

The situation is a Trojan horse for publishers and creators whose labor trained the models, and the algorithm acts as a carnival mirror that magnifies the body type and cadence it prefers.

Can comedians protect their material?

Short answer: not reliably yet. Reporting tools are slow and ambiguous; copyright law is a messy fit for short-form jokes. Comedians like Carter said they stopped posting new bits online because the risk-reward calculation changed. You can register jokes, pursue takedowns, and call out theft publicly — Olivia Carter and George Brett Olson did — but enforcement lags innovation.

If you’re a creator, consider shifting how you publish: exclusive tickets, live streams with proof of performance dates, or membership tiers that give fans verified access. A $10 (€9) subscription can be replicated by a bot, but a live, dated performance is harder to convincingly fake in real time.

At the crossroads: public trust, policy, and the next wave of AI tools

This moment is not just about jokes. It’s a test of platform governance, creator rights, and investor signals. Moore’s casual praise functions as an authority cue that influences startup behavior. Investors and founders will chase whatever the metrics reward — if virality converts to revenue, expect more of the same.

Microsoft, OpenAI, Tavus, a16z, OnlyFans, Fanvue, Instagram — they’re all actors in a system that currently values engagement over provenance. That imbalance creates incentives for bad actors and difficult choices for honest creators.

I’ll leave you with this: the tools that can simulate a smile can also simulate your work. What will you believe tomorrow when the platform serves you a perfect, synthetic comic with a direct link to a paid page?