Italian brainrot: what it is, where it came from, and how to make your own AI brainrot video

Italian brainrot is a meme format of AI-made hybrid creatures with fake-Italian names, narrated by a robotic voice that reads a nonsense rhyme. Hook Studio is an AI reel maker that helps creators turn an idea like this into a short vertical video for TikTok and Instagram. This guide covers the history first, then the recipe.

Italian brainrot is a wave of AI-generated memes that started in early 2025. Each one shows an absurd hybrid creature, such as a shark in sneakers, with an Italian-sounding name and a synthetic voice that reads a rhyme. Despite the name, it did not start in Italy, and the "Italian" is mostly a sound: invented words that end in -ini, -ello, or -ino.

Everything below is how the trend stood as of October 2026, based on Wikipedia, Know Your Meme, and press coverage. The exact origin is disputed, so treat the dates as the earliest known posts, not as an official history.

The first widely seen character was Tralalero Tralala, in January 2025. The chant behind it is older: it comes from internet memes of Dwayne Johnson rhyming about absurd things in late 2023.

• Early January 2025: a TikTok video of a shark in sneakers, Tralalero Tralala, is often named as the first. Sources disagree on who posted it first and on the exact design.
• January 13, 2025: the earliest known notable video using the AI image, per Know Your Meme.
• February 2025: Tung Tung Tung Sahur, a wooden log with a bat. Its name comes from Indonesia: the sound imitates the drum that wakes people for the pre-dawn Ramadan meal.
• Spring 2025: Bombardiro Crocodilo, Ballerina Cappuccina, Lirili Larila, and Brr Brr Patapim spread.
• Mid 2025: trading cards and sticker albums appear in Italy, and the Roblox game Steal a Brainrot becomes one of the platform's most popular games.
• 2026: Wikipedia reports Tung Tung Tung Sahur and Ballerina Cappuccina skins in Fortnite in April 2026, and a legal dispute over who owns an AI-made character.

Most characters follow one rule: take two things that do not belong together and give the result a name that rhymes. These are the ones people search for the most.

• Tralalero Tralala: a shark in sneakers.
• Tung Tung Tung Sahur: a wooden log holding a bat.
• Bombardiro Crocodilo: a crocodile head on a bomber plane. Its narration drew criticism for a reference to bombing in Gaza.
• Ballerina Cappuccina: a ballerina with a cappuccino cup for a head.
• Trippi Troppi: a cat with a shrimp body.
• Chimpanzini Bananini: a chimpanzee with a banana body.
• Lirili Larila: a cactus-elephant in sandals.

It spread because anyone could make one in an afternoon. You need no camera, no actor, and no editing skill. An image model draws the creature, a tool animates it, and text-to-speech reads the rhyme. One analysis of the trend (Network Cultures, September 2025) also credits the fake-Italian language: it sounds like something you almost understand, which is part of the joke.

The same analysis noted that by late 2025 attention was moving to more realistic AI clips, and that the trend had turned commercial and child-friendly. Fan sites in 2026 still list new characters every month, but I found no independent measure of whether it is growing or fading. Search interest is still large: Italian brainrot gets about 201,000 searches a month in the US, and the Spanish term "brainrot italiano" about 18,100 in Mexico. For more formats like this, see how people make viral AI videos and AI cat videos.

An AI brainrot video is three layers: a picture, a short motion, and a voice. Nothing else is needed, which is why the format is cheap.

• The picture: an image model draws the hybrid creature from a text description.
• The motion: a video model turns that picture into a 4 to 10 second clip, often a slow turn, a dance, or a stare.
• The voice: a text-to-speech voice reads a rhyme with the character's name in it. Network Cultures reports many early videos used the same default ElevenLabs voice.
• The finish: captions, a sound from the app's library, and a cut to fit 9:16.

Make a new creature. The famous characters are the most crowded part of the format, a legal dispute over who owns an AI-made character was still open in 2026 per Wikipedia, and a new character is the only thing that is yours. Mix two objects, give it a rhyming name, and write the name before you draw anything. Hook Studio also never copies a celebrity's face, so keep real people out of it.

The recipe is five steps and about 123 credits before the render. You confirm the plan and see each price before anything runs, and a failed clip refunds its credits.

1. Make the still

Tell the agent the creature in one line, like "a teapot with a flamingo's legs, wearing a tiny crown, kitchen counter". A still costs 11 credits. A still is cheap on purpose: you want to see the creature before you pay for motion.

2. Edit the still until it is right

Steer in short lines: "rounder", "less messy", "night kitchen". An edit keeps the creature and the scene and changes only what you asked. Each edit is a new still, 11 credits.

3. Turn the still into a clip

Only now make the clip. Kling 3 at 1080p costs 96 credits for 5 seconds (77 for 4, 153 for 8). Ask for simple motion: a slow turn, a hop, a stare at the camera. Simple motion fails less. Clips come back silent. You can read more on Kling 3 and 4 in Hook Studio.

4. Write the voice line

A voice line reads the text you write, in one stock voice, for 5 credits. Write a short rhyme with the creature's name, and invented words that look Italian: "Teapotto Flamingini, tiki tiki spaghettini". The voice reads the text as written. Hook Studio does not promise a particular accent, and it cannot clone anyone's voice, so spell the words the way you want them said and listen to the result.

5. Put the voice over the clip and render

Use the free replace-audio edit to lay the voice line on the silent clip. You do not need lip sync, because the creature is not a person talking: the voice is narration. Then the agent renders the reel for a few credits, charged by measured time, and you download it to post. Direct posting is not available yet, and every download comes with a reminder to turn on the AI label, which TikTok and Instagram ask for on realistic AI content.

One 5 second brainrot video is about 123 credits plus the render, about $1.07 on the Concept plan. Prices leave out the realism checks, a few credits each. Your first $5 costs $3, 500 credits, which covers about four tries at a single still and clip.

• No guaranteed accent. The voice reads your text; it does not act Italian.
• No music or ambient sound from the clip. Add a sound from the TikTok or Instagram library when you post.
• Clips are 4 to 10 seconds. A longer video is several shots joined.
• No promise of views. The format is crowded; a new character and a good rhyme help, nothing guarantees.

If you want the general workflow, see the AI reel maker page. Hook Studio is not affiliated with Kling, TikTok, or Instagram.