Sora 2 for brands: what’s left after the shutdown and what to use now
OpenAI shut Sora 2 down in 2026. The dates, the reported reasons, what to do with videos already generated and which engines to use instead in November 2026.
Sora 2 can no longer be used: OpenAI announced its closure on 24 March 2026, shut down the app on 26 April and the API on 24 September 2026. As of November 2026 there is no country in which a brand can generate video with Sora 2, and OpenAI has not named a replacement. What remains are the videos already made, with their watermarks and metadata, and some useful lessons about rights, consent and depending on a single supplier.
If you had planned a campaign around Sora 2, or keep finding guides that still recommend it, this is what happened, what to do with existing content and which engines to use instead, with verified dates.
What Sora 2 was and why brands paid attention
Sora 2 was OpenAI’s second video model. The original Sora had arrived in December 2024 for ChatGPT Plus and Pro subscribers in the US and Canada. Sora 2 was unveiled on 30 September 2025 together with an iOS app, opened at first to selected users in the same two countries, as Wikipedia’s entry on Sora records. The Android app followed about two months later.
The news was not just the model. The app had social network features, to the point that outlets such as Wired compared it to TikTok. For a brand it was a strong idea: generation and distribution in the same place, with an audience constantly making and resharing videos.
That very success brought the problems to the surface: protected characters generated without permission, deepfakes of people who had died, watermarks removed within days. These are the issues that still matter today, whatever the engine.
Why OpenAI shut Sora 2 down
The whole arc lasted less than a year. These are the key dates, according to Wikipedia and the sources it cites.
| Date | What happened |
|---|---|
| 30 September 2025 | Sora 2 and the iOS app launch |
| 11 December 2025 | Disney deal: a $1 billion investment and a three-year character licence |
| 24 March 2026 | OpenAI announces on X that the app and API will close |
| Late March 2026 | Disney exits the deal |
| 26 April 2026 | The app shuts down |
| 24 September 2026 | The API shuts down |
In its shutdown notice OpenAI gave no specific reason. Press accounts point to a shortage of computing capacity, cost pressure and a shift towards enterprise products. According to TechCrunch, as cited by Wikipedia, users peaked at around one million before falling below 500,000, and the service cost an estimated $1 million a day to run.
The entry on OpenAI links Disney’s exit directly to Sora’s closure. The details of the deal, and what it means for character licensing, are in our piece on Disney, OpenAI and Sora.
Rights and consent: what Sora 2 taught us
At launch Sora 2 generated copyrighted characters by default: it was up to rights holders to ask OpenAI to exclude them. On 3 October 2025 OpenAI promised more granular control. On 6 October the Motion Picture Association criticised the approach, and in the same month Japan’s Content Overseas Distribution Association, whose members include Studio Ghibli and Square Enix, demanded that OpenAI stop using its members' work.
On the likeness side, the estates of famous people who had died protested about deepfakes. OpenAI blocked videos of Martin Luther King Jr. and gave estates the option to opt out. When the Disney deal arrived in December, SAG-AFTRA, the Writers Guild and the Animation Guild set out their concerns in an open letter the very next day.
The lesson for a brand is simple. An opt-out system is not a licence: the fact that a platform lets you generate a character or a face does not mean you can use it in a campaign. Rights are secured in writing, before anything is generated.
Already have videos made with Sora 2? What to do
Videos generated before the closure are still your files, but they need handling with care. According to Wikipedia they carry a visible, moving watermark and C2PA metadata identifying them as AI-generated.
- Archive the originals. Keep the files exactly as downloaded. Content Credentials only survive if the file is copied and published unaltered, as the entry on the Content Authenticity Initiative explains.
- Don’t remove the watermark. Tools to strip it appeared a week after launch, but a brand that hides a video’s AI origin is taking a risk. The transparency obligations are covered in our guide to the AI Act.
- Reread the terms. Your usage rights depend on the terms of service you accepted when you generated the videos: check them before reusing anything in a new campaign.
- Document everything. Prompts, references, dates and versions are what you need to rebuild a scene on another engine if the client asks for a variant.
What to use instead of Sora 2 in November 2026
Sora no longer appears on the platforms we checked: neither in Adobe’s partner model list, updated on 23 October 2026, nor among Higgsfield’s video models. There is no single replacement: it depends on what you needed it for.
| Need | Engine | Why |
|---|---|---|
| A line with lip sync | Veo 3.1 | Native audio, 4 to 8-second clips |
| A scene with several cuts | Kling 3.0 | Up to 6 shots in 15 seconds |
| A long take | Kling 4.0, Seedance 2.5 | Up to 30 seconds in one pass |
| Editing existing video | Gemini Omni Flash | Edits and extends existing clips |
The practical details are in our guides to Veo 3.1, Kling 3.0 and Seedance. The technical sources are fal’s pages on Veo 3.1, Seedance 2.5 and Gemini Omni Flash, Kling’s multi-shot guide and the Kling 4.0 announcement of 15 October 2026.
The lesson: a project cannot depend on a single engine
Anyone who had built a format, a series or an entire workflow on Sora 2 had to start over within months. It is not an isolated case: models change version, name and terms at a steady pace, and a supplier can change strategy from one quarter to the next.
The defence is to keep the project independent of the engine. Storyboards, keyframes, character references, prompts and rights paperwork are the real assets, because they travel from one model to the next. The engine is chosen shot by shot, and it can be replaced.
That is how we work in the production act: the right engine for every scene, client approval at every step and rights written into the proposal before work starts. If Sora disappears, the film survives.
Frequently asked questions
Is Sora 2 still available?
No. OpenAI announced its closure on 24 March 2026: the app was shut down on 26 April 2026 and the API on 24 September 2026. As of November 2026 you cannot generate new videos with Sora 2.
Can I still use videos made with Sora 2 in a campaign?
That depends on the terms of service you accepted when you generated them and on what the video contains, such as real faces or protected characters. Keep the original files with their watermark and metadata, and check the rights before reusing them.
What is the best alternative to Sora 2?
There isn’t just one. As of November 2026, Veo 3.1 suits lines with lip sync, Kling 3.0 scenes with several cuts, Kling 4.0 and Seedance 2.5 takes of up to 30 seconds, and Gemini Omni Flash editing existing video.
Sources
- Sora (text-to-video model) · Wikipedia, 2026-11
- OpenAI · Wikipedia, 2026-11
- Text-to-video · Wikipedia, 2026-11
- Content Authenticity Initiative · Wikipedia, 2026-11
- Partner models in Adobe products · Adobe Help Center, 2026-10-23
- AI Video Generator: Kling, Veo, Seedance and more · Higgsfield, 2026-11
- Veo 3.1 | Text to Video, Image to Video, Reference to Video, Extend Video · fal, 2026-11
- Seedance 2.5 on fal · fal, 2026-10-21
- Google models on fal (Gemini Omni Flash 1.2) · fal, 2026-11
- Kling VIDEO 3.0 Multi-Shot: Create Structured Cinematic Sequences · Kling AI, 2026-07-28
- Kling AI Officially Launches Next-Generation Kling 4.0, Raising the Bar for Professional-Grade Video Models · Kuaishou Technology via Bernama, 2026-10-19
Keep reading.
Kling 3.0: a practical guide to multi-shot, audio, Omni and limits
Multi-shot, native audio, Omni elements, 4K and costs: how Kling 3.0 works in November 2026, what it can’t do, and when it still beats Kling 4.0.
Veo 3.1: a practical guide to audio, references, Flow and SynthID
Native audio, reference images, first and last frames, extension, access and limits: how to use Google Veo 3.1 in November 2026, and when to switch to Gemini Omni.
AI Christmas ads in 2026: who uses AI, who refuses it and why
Prada draws by hand, Levi’s goes back to claymation, Dior faces AI accusations: Christmas 2026 shows what audiences punish and accept from AI.