ByteDance has embedded CapCut into ChatGPT.

On October 24, at the Audrey Irmas Pavilion in Los Angeles.

CapCut, the international version of Jianying, announced at its global creator event CapCut World LA that its video editing and generation capabilities will soon be integrated into ChatGPT. In the official demonstration, the user uploaded 5 football videos and put forward a simple request: edit them into a 12-second vertical highlight reel with the title "GOAL!".

ChatGPT then generated a draft video. The user continued to submit modification requests in natural language, and CapCut adjusted the order of materials, subtitles and timeline. After the rough cut was completed, the project can also be imported back to the CapCut web editor for further effect addition, subtitle configuration, fine-tuning and release.

The official CapCut account described this product as a new plugin that will soon be launched on ChatGPT; the announcement video shows that the open beta program will cover mobile, desktop and web platforms in November.

The AI 2.0 Evolution of CapCut

This is not the first time CapCut has attempted to integrate with ChatGPT.

In July 2023, ByteDance once released a ChatGPT plugin for CapCut. Users only need to enter a sentence, and the plugin can further generate scripts, voiceovers, background music, subtitles and material combinations, and present the results for users to preview and edit.

That generation of products solved the problem of "creating a video from scratch".

It lowered the threshold for video production, but still worked more like an automatic generator: users provided the theme, the system output a preliminary result, and actual modifications still needed to be completed on CapCut.

The new version in 2026 has a different set of tasks.

Users can directly upload their own video materials and ask the system to understand, filter and organize these contents. It is no longer limited to generating content from a blank canvas, but can participate in real editing tasks.

The difference between the two can be summarized as:

In 2023, AI helps you conceive a video.

In 2026, AI starts to help you process a large batch of materials.

The former lowers the creation threshold, while the latter truly integrates into the content production workflow.

In the past, editing a video usually went through several steps.

First open the editing software, import the materials, then manually browse the timeline, select clips, adjust the order, add text, match music, and finally export versions required by different platforms.

The software provides a large number of functions, but users need to know where the functions are and in what order to complete the tasks.

The change demonstrated by CapCut this time is that users do not need to understand the operation logic of the editing software first.

They only need to clearly describe the desired result:

"Edit these clips into a 12-second vertical football highlight reel with a fast pace and the title GOAL!."

The conversational model is responsible for understanding the task, and CapCut is responsible for calling materials, organizing the timeline and generating the rough cut.

There is an obvious change in this:

The focus of previous AI video products was "helping you generate a video", while the current focus has shifted to "helping you turn the materials you have on hand into a usable video".

Generating a seemingly complete video is no longer the most difficult part. What really consumes time is selecting which clips to use, how to sort them, where to place subtitles, whether the rhythm is appropriate, and how to adapt the same content to different platforms.

CapCut's integration into ChatGPT this time is precisely targeting these repetitive but high-frequency production links.

What Exactly Is the Moat of Video Editing?

CapCut's core advantage has never been limited to its editing functions.

It is equipped with a huge library of templates, music, special effects, stickers, transitions and massive creator-generated content. Once users enter CapCut, they often complete material selection, video production, modification and release on the same platform.

However, the emergence of generative AI is changing the order in which users access tools.

Users may first discuss topics, write scripts, and plan shots in ChatGPT, and then decide what materials and editing methods they need.

If this step takes place in ChatGPT, CapCut will face a real risk: it still has the most powerful editing capabilities, but it may no longer capture the user's initial creative intention.

Users may have completed content planning in the chat window, and CapCut will only be treated as an "export tool" at the final stage.

Therefore, CapCut needs to position itself at the very beginning of the creation process.

The official manifesto at CapCut World has clearly stated this positioning change: CapCut is no longer just a video editing tool, creators can also carry out design, generation, directing and collaboration on it.

The so-called plugin is exactly the way to embed CapCut's capabilities into the scenario where creative intentions are formed.

Users first state what they want to do, and CapCut then converts these intentions into projects, materials and timelines.

This is a new division of labor:

- ChatGPT is responsible for understanding goals, supplementing requirements and organizing tasks;

- CapCut is responsible for processing materials, timelines, special effects and finalizing videos;

- Users are responsible for confirming the direction, modifying the results and final release.

What CapCut gains is not a single call, but the entire process of users from "having an idea" to "getting a finalized video".

From a practical work perspective, the first things to be automated will not be aesthetics and judgment, but repetitive operations.

For example:

- Filtering usable clips from a large number of materials, arranging shots according to the script or theme, deleting pauses and repeated content, converting horizontal videos into vertical versions; automatically adding subtitles and multi-language versions, batch generating versions with different durations and aspect ratios for different platforms, searching for music, transitions and special effects that match the emotion and rhythm in the material library.

These tasks originally required editors to drag the timeline continuously, but they do not necessarily constitute the core of creation.

The parts that are really difficult to replace still include:

- Whether the finalized video is accurate, natural and credible.

In other words, the value of video editors will shift from "being familiar with software buttons" to "defining results and judging results".