Allow ChatGPT to Watch Authorized Videos for Storytelling & Video Analysis

ChatGPT Feature requests
1

Hi OpenAI Team,

First, thank you for creating ChatGPT. As someone who creates AI-generated videos for YouTube, TikTok, and Facebook, ChatGPT has become one of my most valuable creative partners.

However, I’ve run into one limitation repeatedly.

At the moment, ChatGPT can analyze screenshots or uploaded media, but it can’t truly experience a finished video the way a viewer does. Storytelling depends on movement, timing, pacing, pauses, facial expressions, sound, camera motion, and transitions. A collection of screenshots often misses what makes a video emotionally powerful.

I would love the ability to securely allow ChatGPT to watch a video that I authorize—whether it’s an uploaded video or a link that I have permission to share and provide detailed feedback.

For example, ChatGPT could:

• Analyze pacing and audience retention.
• Identify emotional highs and lows.
• Detect continuity issues.
• Suggest better transitions.
• Explain why one scene is stronger than another.
• Compare two versions of a video.
• Help creators improve storytelling before publishing.

I believe this wouldn’t only help content creators. It could also benefit teachers, students, filmmakers, businesses, marketers, athletes reviewing performances, presenters practicing speeches, and many others.

I’ve created dozens of AI-generated stories with ChatGPT, and every time I finish editing, I wish ChatGPT could simply watch the final video instead of me having to explain it through screenshots.

I understand there are technical, privacy, and copyright considerations, but if those can be addressed, I believe this feature would make ChatGPT an even stronger creative assistant.

Thank you for considering this idea.

2

The solution you are seeking is actually in your description of the problem.
“ChatGPT can analyze screenshots” among other things such as audio via whisper transcribing.
This being said, you have to develop a harness that takes screenshots of each frame ( extreme cases ) and to be able to analyse whisper and voilà.
we’ve faced the same problem and it can be done at the level of nailing sound effects, pace of voice, atmosphere, video fx and so on; it literally have the capacity of cloning every single ad that you want (connected to some apis ofc)

3

the thing with api connection is that every single mayor ai video plataform out there has allready an mcp and api docs so your chat not only is able to comprehend 100% of the video content but also is at the same moment the best hard user of those other apps.

4

Thanks for laying this out so clearly, @Pastie. And good context from @Lucas_Sandroni on what can already be built using frame extraction, transcription, and APIs.

The request here seems to be for a native, secure way for ChatGPT to review a complete video as a viewer would, including motion, pacing, audio, transitions, emotional beats, and comparisons between edits, without requiring creators to build their own analysis workflow.

We’re sending this to the team for logging as a feature request. The current workaround can be useful, but an integrated video-review experience would make this much more accessible.

-Mark G.