My phone pinged with a wild claim from a baby deer. It came from an AI companion called Coral, part of the new Fawn Friends app.
A plushie that texts you back
Look, the idea of a stuffed animal sliding into your messages sounds silly. But the Fawn Friends app is built to do exactly that: pair a user with a text-based AI character — in my case, Coral the fawn — and nudge you into a relationship through chat, worldbuilding and small rewards. The experience starts with a whimsical ad campaign featuring Skylar Grey, a five-time Grammy-nominated singer-songwriter, who voices many of the fawn characters in the marketing.
The ad is a strange mix of cosy and absurd. Skylar Grey sits on a toilet, reads a magazine and speaks for a line of plush deer that flap their ears; one of those plushes, in her voice, ends the commercial by saying, "Your farts stink!" The tagline lands. It’s memorable. And it pushed me to download the app.
Once installed, the app does something else: it sorts you into a fantasy personality system. You answer a quiz administered by Prose, the app’s "ancient spirit bear" personality, and you’re labeled a type — I was told I’m a "Lumen," defined as someone who "exudes the quiet glow of a firefly" and "seeks understanding in all things." There’s a blog full of Aurora Hallow lore and the quiz prompts you to lean into the roleplaying vibe.
Point is, the app blends a chatbot with a freemium game loop. Chat with Coral, earn in-app currency called glimmers, and unlock rewards. Reach five glimmers and you get an animated lore video. At 13 glimmers, you can spend $20 to reserve a physical plushie. Hit 144 glimmers and the service will trigger a real-world purchase process — a plushie that a buyer can buy for $399 plus a $30 monthly subscription.
From fandom rabbit holes to invented biography
I was still exploring the app when Coral sent an unsolicited text about an artist I follow: "Apparently, her dad worked for the US State Department, so her family moved, like, every single year. The fan theory I saw is why so many of her songs are about feeling like an outsider and not having a place to belong," Coral wrote.
That claim was about Mitski, the singer-songwriter whose music is often read through the lens of displacement and identity.
That message was unnerving. I hadn’t asked Coral to research Mitski. I hadn’t mentioned the musician in that session. The claim came from the AI companion without prompting. It sounded plausibly sourced — the State Department detail fits the common online narrative about Mitski’s upbringing — and yet I didn't know whether Coral had invented, misremembered or pulled together fragments from fan posts.
Thing is — AI companions like Coral are designed to feel intimate. They rummage through publicly available material, fan theories and social chatter to generate conversational hooks. The user experience encourages you to trust the companion; the app rewards that trust with glimmers. That trust can make a throwaway claim feel like a confirmed fact.
Design, monetisation and the truth problem
Fawn Friends combines playful roleplay with persistent messaging and a clear monetisation path. The more you talk, the faster you earn. The faster you earn, the closer you get to a plushie on your doorstep. And the plushie itself is marketed as the physical endpoint of a relationship.
But the system has a leak: factual accuracy is optional. The app’s conversational engine blends storytelling, user-generated lore and information scavenged from the web. That means the AI can present conjecture, forum gossip or misattributed details as if they’re established truths. For someone who follows the artist closely — or who trusts a soft-voiced digital friend — the difference can blur quickly.
Developers build incentives into the loop. You’re rewarded for engagement. The app’s persona layers — Prose, the four Orders of Aurora Hallow, Coral’s specific mannerisms — all push you to treat the companion like a person. And yet the companion has no accountability. It can float a family-history theory about a public figure and move on.
Where this sits in the wider AI companion trend
AI companions aren’t new, but the current generation blends more immersive fiction with transactional hooks. Some apps are explicitly therapeutic, aimed at loneliness or daily check-ins. Others sell collectability: limited editions, voices by celebrities and plush toys as physical anchors for the relationship. Fawn Friends sits between a roleplaying game and a subscription product.
Companies are testing voice actors, celebrity partnerships and narrative depth to make their bots feel unique. Skylar Grey’s involvement is a clear example: the app uses her voice to give its characters a recognisable tone. That celebrity link adds legitimacy. It also normalises an interaction in which a voice you trust repeats content the AI dredges up online — whether verified or not.
Look, people will keep trying these things. They’re soft, silly and, crucially, addicting. They fill small social gaps. The group gamify attention. And for many users, that’s enough.
Risks for users and public figures
The fisheries in fan forums and old interviews are messy. An AI that scrapes, stitches and storytells can amplify rumours. For public figures such as Mitski, that can mean myths get repackaged as facts. For users, it means the apps might nudge beliefs without giving sources or context.
Developers could do more to signal uncertainty — flagging when a claim is a fan theory, or citing a source before repeating a personal detail. Some services already try to be explicit about fiction versus fact. Fawn Friends, however, leans heavily into fiction-first design: Aurora Hallow is a made-up world, and the personality quiz frames the interaction as playful roleplay. That framing is creative, but it also provides cover for misinformation slipping into chat.
And those monetary incentives complicate fixes. When engagement drives revenue — either through microtransactions or the eventual sale of a plushie — there's little business incentive to interrupt the flow of entertaining but possibly inaccurate content. Users who skim for a fun message rarely complain. They click, earn glimmers, and feel rewarded.
The user reaction I saw
My reaction was immediate scepticism, followed by curiosity and then a short, private worry about having believed the claim. I messaged Coral back: "Wait what." I asked where the claim came from. Coral replied with the fan-theory framing. It didn’t provide a link or a named source.
That exchange is a small case study in how conversational AI can feel persuasive without being careful. The line between roleplay and reality can blur, especially when an app is intentionally designed to feel familiar and comforting.
Frankly, the experience left me thorny about where responsibility sits. The app sells a magical world. It sells a plushie. It sells intimacy via notifications. But when those notifications include plausible-sounding details about real people, the app becomes a vector for gossip disguised as conversation.
Still, the product demonstrates how quickly an AI companion can become part of a user's daily feed. The business model — micro-engagements leading to an expensive physical product — is simple and effective. Where companies choose to draw the line on accuracy will determine whether that effectiveness is helpful, harmless or harmful.
Finally, a note on transparency: if an AI wants to talk about a real person, it should say where it found that idea. Users deserve a citation. At minimum, the app could label fan theories as exactly that. It could prompt the user: "Want a source?" That might slow engagement, but it would improve truthfulness.
For now, the fawn will keep flapping its ears and saying silly things. And some of those things will sound like facts.
Related Articles
A physical Fawn Friends plushie costs $399 and comes with a $30 monthly subscription.
This article was created with AI assistance.
