OpenAI has officially ended its Sora text-to-video application, with the service shutting down on April 26, 2026, after just 16 months of public availability. The company has also announced that the Sora API will be discontinued on September 24, 2026, marking the complete termination of the project.
Sora first launched for ChatGPT Plus and Pro subscribers in the United States and Canada on December 15, 2024. The tool allowed users to generate short video clips from text prompts and extend existing short videos. OpenAI had previewed sample outputs in February 2024, so the model’s capabilities were known before its public release.
From the outset, the project faced a significant challenge. By default, Sora trained on copyrighted material and incorporated that content into the videos it generated.
Copyright holders were required to opt out of this arrangement, a policy that drew immediate criticism from artists, studios, and stock footage companies. OpenAI’s response was to embed a visible, moving digital watermark in every Sora video. The watermark was intended to allow viewers to identify clips as AI-generated.
However, this safeguard lasted less than one week. The turning point came with the launch of Sora 2 at the end of September 2025.
This major update introduced social media features and expanded access to a broader group of select users in the U.S. and Canada. Almost immediately, third-party programs emerged that could strip the watermark from videos, rendering the protection useless. Without a reliable method to mark its output, Sora became a liability. The technology could produce convincing fakes with no identifiable fingerprint.
While Sora 2 demonstrated OpenAI’s commitment to improving the technology with new social features and user interactions, the watermark problem remained unresolved. No second-generation safeguard replaced the broken system.
The app continued operating for seven months after the Sora 2 release before shutting down. The API will follow in September 2026, after which no part of the Sora brand will remain. The timeline of Sora’s existence is notably compressed: a preview in February 2024, public release in December 2024, a major update in September 2025, and shutdown in April 2026.
The entire arc from debut to termination spanned approximately two years and two months. OpenAI has not publicly commented on the reasons for Sora’s discontinuation.
However, the sequence of events points to several factors: a watermark that failed within a week, a copyright system that required rights holders to opt out of having their work used by default, a social media update that expanded the user base without addressing the core vulnerability, and third-party tools that erased the only identifier the company had implemented. Sora was a major product release from a leading AI company, with paying subscribers on two tiers and a second generation featuring new capabilities. Despite this, it could not maintain a working content-authentication system for more than a single week.
The September API sunset will close the project entirely, with no plans for Sora 3, a watermark fix, or an opt-in copyright framework. The project lasted 16 months in public service, with one week of working protection.
{“@context”: “https://schema.org”, “@type”: “NewsArticle”, “headline”: “OpenAI Shuts Down Sora Video App After Just 16 Months”, “description”: “OpenAI has officially ended its Sora text-to-video application, with the service shutting down on April 26, 2026, after just 16 months of public availability. T”, “datePublished”: “2026-06-25T09:30:00+00:00”, “author”: {“@type”: “Person”, “name”: “Maria Romano”}, “publisher”: {“@type”: “Organization”, “name”: “InfoPulse Today”, “url”: “https://newsfeed.infopulsetoday.com/”}, “image”: “https://freesuntimes.com/wp-content/uploads/sites/20/2026/05/companion_3411.jpg”}
