The Drake What Did I Miss AI trend is a 15-second clip in which one person raps along to Drake's song What Did I Miss? in a grey cold storage room full of ice blocks. You can make it from a single photo of yourself, with no prompt, and the clip comes with the song's sound already on it.
The page itself answers the basics. This guide goes one step deeper: what the edit format is, why the cold-room clip spread, how the motion reference works, how to keep your face, what happens to the Drake audio on each platform, how to label the result, and what it costs.
Key takeaways
- The trend is a wave of fan edits and AI remakes of one song, not a single dance. The AI version puts you in the cold room for 15 seconds.
- A motion reference clip supplies the camera, timing and gestures; your photo supplies the face, hair, skin tone and build.
- The audio is Drake's recording, so a platform can mute it or limit reach. A licensed sound from the platform's own library is the fallback.
- Label the clip as AI-generated. YouTube and Meta both have a way to do it.
- A 15-second Standard 720p video is 13 credits, and your first video on an account never costs more than 10.
In this guide
- What is the What Did I Miss edit format?
- Why did the cold-room clip take off?
- How does the motion reference work?
- How do I keep my face in the video?
- Which photo works best?
- How is this different from a lip-sync filter?
- Will the Drake audio be muted or flagged?
- How do I label the clip as AI content?
- What does it cost?
- How long will the trend stay relevant?
What is the What Did I Miss edit format?
It is a music trend built on Drake's song What Did I Miss?, not a choreographed dance. Most of the posts are fan edits that cut clips of the artist to the track. The most-viewed ones show a lone man in a navy work jacket rapping and gesturing in a grey cold storage room full of ice blocks, cut with fur coats, red sports cars, a glowing blue tower and quick close-ups.
The song is named in the captions of the top clips, and a CapCut template called What did I miss? circulated from July 2025. In the trend the title works as a hook: someone comes back and asks what changed while they were away. No original creator is confirmed, so treat any claim of "who started it" with caution.
Why did the cold-room clip take off?
We cannot see inside the platforms, so this is an observation, not a measurement. The cold-room clip is easy to read in a second: one figure, one fixed wide camera, a slow push-in, and a room that looks unlike a normal music video. It needs no costume change and no dance skill, only loose gestures that anyone can imitate.
The earliest high-view edit we found is dated 2025-07-05 and has about 6.9 million views. Other edits from June to October 2026 have 3.6 million, 1.5 million and 428 thousand views. That spread of dates matters: this is a long-running edit style, and the AI remake is the new part. It only started appearing in late October 2026.
How does the motion reference work?
The videosi clip is made with ByteDance's Seedance model from your photo plus a short motion reference of the cold-room performance. The reference carries the camera, the timing and the gestures; the model swaps the person and keeps the rest. That is why you get the same room and moves every time instead of a loose guess from a text description.
The 15 seconds follow six beats: hands in pockets, chest taps, pointing at the lens, a hand over the face with a bowed head, a look up, and a low crouch pointing at the floor. You can pick the cold storage room or a dim garage with red sports cars and a fur-collar denim coat. The full prompt stays built in on the Drake What Did I Miss AI page, so you only choose the photo, the setting and the format (9:16, 1:1 or 16:9).
How do I keep my face in the video?
Two things matter. First, the identity lock: your photo is the only source for your face, hair, skin tone and build, and the model is told never to reshape a face toward the rapper. Without that, models tend to keep the original man's braids, beard and jacket and only borrow your clothes.
Second, expect some drift in the hardest beats. Likeness is not guaranteed in the roughly 3-second head-bow and hand-over-face moment, where hair and beards drift most. If your first result is slightly off there, try another photo before you change anything else. Hands in pockets and hands over the face are also where models sometimes add extra fingers or merge a hand with the face, so check those seconds before posting.
Which photo works best?
Use one sharp, evenly lit photo that shows your face and shoulders, with only you in frame. The usual causes of a bad match are a cropped face, heavy shadow, a hat or sunglasses, and two people in the picture. A cold, desaturated room can hide skin tone and face detail, so a photo with even, natural light gives the model more to hold on to.
Only use your own photo or one you have permission for. The tool turns one photo of a single adult into the performance and does not put a real person into another person's likeness. If you want a second person, make a separate clip for each and join them in an editor.
How is this different from a lip-sync filter?
A lip-sync filter keeps your own footage and animates your mouth to the audio. This is a full video generated from a still photo: the room, the camera move and the body language all come from the motion reference, and you never record anything yourself.
The trade-off is that your face is not lip-synced to the lyrics. The person gestures as if rapping along, and the sound is lined up with the moves, but the mouth is not matched word for word. If you need exact lip sync, this is not the right format.
Will the Drake audio be muted or flagged?
It can be. The sound is Drake's recording, so posting it carries the usual music rights risk, and the result depends on the platform and on rights in your region. This is general information, not legal advice.
YouTube's help pages on Content ID and on copyright claims on music describe how it works there. An upload that matches a rights holder's reference file gets a claim, and the owner can block, monetize or track the video, with different outcomes by country. Per those pages, claimed Shorts of 1 to 3 minutes are blocked while a claim is active, without a copyright strike. For audio claims you may be able to mute the claimed song, trim the segment or replace it with audio from the YouTube Audio Library.
If a platform mutes the sound, swap in a licensed track from its own library and keep the visuals. A 15-second clip with a different sound still works as a cold-room performance.
How do I label the clip as AI content?
YouTube asks creators to disclose realistic content that is made or altered with AI, including a realistic scene that did not happen. In YouTube's disclosure guidance, creators select "AI use" under Attributes in YouTube Studio when uploading. The page also says YouTube may label content automatically, and that creators who consistently fail to disclose can face penalties.
Meta takes a different route. In its post on labeling AI-generated content, Meta says it adds an "AI info" label when it detects industry-standard AI indicators or when a user self-reports AI-generated material. Labeled content is not removed for that reason alone. Our advice is simple: tell viewers it is AI in the caption, use the platform's own label where there is one, and never present the clip as real footage of you or of Drake.
What does it cost?
It is not free. A 15-second Standard 720p video is 13 credits, and your first video on an account never costs more than 10 credits. The Single pack is $6.99 for 10 credits, so it covers a first video. Better models and higher resolutions cost more, and a failed render returns its credits.
Doing it yourself with a text prompt in another tool often takes several tries, because the model keeps the rapper's face or loses the slow push-in, and every try costs time and credits. The motion reference is what removes most of that guesswork. You can start on the Drake What Did I Miss AI page.
How long will the trend stay relevant?
Nobody can promise that. Fan edits of the song have circulated since July 2025, and as of October 2026 new edits are still being posted. The AI remake format is just starting, so we mark it as emerging rather than established.
Joining early costs one photo and 15 seconds. If the trend fades, the clip still works as a short rap-performance video with your face in it, and you can swap the sound for something licensed whenever you like.
