Can you use AI to make videos from images?

Short answer: you’ll need a different tool. ChatGPT can still help you prepare the image and plan the clip, but as of 11 October 2026 it can’t make the video itself.

What ChatGPT can do with a still image today

OpenAI3’s ChatGPT capabilities overview, updated in October 2026, lists what ChatGPT does with images. It can analyse a photo you upload, generate new images from a text prompt and edit an image when you describe the change in plain words. The page lists no video generation.

OpenAI’s video product was Sora, and it has gone. OpenAI’s help centre says the Sora web and app experiences were discontinued on 26 April 2026 and the Sora API on 24 September 2026. If you made videos in Sora, that help page explains how to export them.

So use ChatGPT for the jobs around the video: cleaning up the still, describing the motion you want and planning a sequence of shots. If OpenAI brings video back, its help centre and release notes will say so first, so check there before you trust any article on this, including this one.

Three ways to turn images into video

1. An AI assistant with video built in. Google’s Gemini app is one. Its help page says you can make a video from a text prompt plus up to five images, using its Gemini Omni model. On a personal account you need a Google AI plan, and on a work or school account you need a qualifying Workspace licence. The feature isn’t open to under-18s. The page says uploading a video to edit isn’t available in the UK, but it lists no such restriction for images. Google also says you must have the rights to any image you upload. Prices and limits aren’t on that page, so check them before you pay.

2. A dedicated image-to-video tool. These are built to animate a single still. Before you upload anything, read the tool’s terms on what it can do with your images and the finished clip, and check it works in the UK. For how these tools fit into a wider workflow, see making YouTube videos with AI.

3. A slideshow-style edit. This one gets overlooked. There’s no AI generation at all: you put your stills in an ordinary video editor, add slow pans and zooms, drop in music and export. It’s predictable, it’s yours, and it avoids most of the likeness and labelling questions that generated video raises. If your photos are already good, it can look better than a wobbly AI animation.

If the source image is rough, fix that first. Tidying it up with AI photo editing gives any generator more to work with.

How to do it well

Start with a photo you took yourself, or one you have permission to use. A sharp, well-lit image with a clear subject gives any tool a better starting point than a blurry crop.

Keep the ask small. Say what should move and how: clouds drifting, the camera easing in, leaves shifting. Vague prompts get vague results, and ambitious ones tend to produce something odd. Start modest and build up. ChatGPT is good at turning your rough idea into a tight, specific motion prompt.

Expect to try more than once. Generated video is rarely right first time. Save the version you like and note the prompt that produced it, because you will forget.

Watch the result twice before you share it: once normally, and once looking at the edges, where faces, hands and text tend to go strange. If something looks off, it probably is.

Decide up front where the clip is going. A video for a private family chat is a different matter from one you post publicly or use for a business, and the rules below bite harder the further it travels.

Rules and risks

The source photo. The Intellectual Property Office7 says whoever creates an image is generally the first owner of the copyright. Photos a friend or relative took need their permission, and a professional photographer usually keeps the copyright in photos you commissioned, such as wedding pictures. The IPO also says adapting someone else’s image without permission will usually infringe their copyright, so running it through an AI tool doesn’t make it yours.

The output. Section 9(3) of the Copyright, Designs and Patents Act 1988 says the author of a computer-generated work is the person who made the arrangements necessary to create it. How that applies to a particular AI clip isn’t settled by the Act or the IPO’s guidance, and the tool’s own terms also affect what you can do with the result. For the wider picture, see whether AI art is illegal.

Photos of real people. The ICO says that if someone can be recognised from a photograph, it’s usually their personal data. Photos kept for personal use, such as a family album or a private account seen only by friends and family, fall outside data protection law. Posting to a public account is likely to go beyond personal use, and the ICO says you then need to think about whether the people shown would reasonably expect it. The tools set rules too. OpenAI’s usage policies ban using someone’s likeness, including their photorealistic image or voice, without consent in ways that could confuse authenticity. Google’s Gemini help says not to violate others’ privacy rights. The simple habit is to animate only people who know about it and are happy with it. Adding a cloned voice raises its own problems, covered in copying someone’s voice.

Intimate deepfakes are a crime. Section 66B of the Sexual Offences Act 2003, inserted by the Online Safety Act 2023, makes it an offence in England and Wales to intentionally share a photograph or film that shows, or appears to show, another person in an intimate state without their consent. Threatening to share one is also an offence.

Labelling. YouTube requires creators to disclose realistic AI content, including content that makes a real person appear to say or do something they didn’t, or that shows a realistic scene that never happened. You do this with the ‘AI use’ setting in YouTube Studio when you upload. Creators who consistently fail to disclose can have a label applied for them, content removed or be suspended from the YouTube Partner Program. Clearly unrealistic content and minor edits don’t need a label. And if your source image carries a watermark, leave it alone: removing watermarks explains why.

This is general information, not legal advice.