OpenAI has announced the shutdown of its social media application, Sora, a short-form video platform powered by artificial intelligence that garnered significant attention and concern since its launch in September of the previous year. The company conveyed its decision in a brief message, expressing disappointment for users and promising future details on preserving created content. Sora was initially conceived as an AI-first competitor to popular platforms like TikTok, YouTube, and Meta’s Instagram and Facebook, aiming to capture the burgeoning market for short-form video content and its associated advertising revenue.

However, Sora quickly became a flashpoint for controversy, raising alarms within Hollywood and among various advocacy groups, academics, and experts. Concerns revolved around the potential for users to generate realistic deepfakes, nonconsensual images, and a general proliferation of what was termed “AI slop” from simple text prompts. OpenAI struggled to implement effective guardrails, leading to instances where users created outlandish deepfakes of public figures such as Michael Jackson, Martin Luther King Jr., Mister Rogers, and Sam Altman. The ease with which these safeguards could be bypassed prompted outcries from family estates and actors’ unions, as well as the daughters of deceased public figures like Martin Luther King Jr. and Robin Williams, who publicly requested an end to videos featuring their fathers.

The app’s flagship feature, initially named “cameos,” allowed individuals to scan their faces to create realistic deepfakes of themselves, which could then be used in publicly viewable videos. This feature itself faced legal challenges, with the company Cameo prevailing in a dispute over the name, forcing OpenAI to rename it to “characters.” Beyond public figures, users also leveraged the platform to generate copyrighted characters engaging in unusual activities, such as Mario smoking weed or Pikachu doing ASMR, further highlighting the moderation difficulties and potential legal liabilities for OpenAI.

In a surprising development, Disney, a company notoriously protective of its intellectual property, had entered into a significant licensing deal with OpenAI, involving a reported $1 billion investment. This collaboration would have allowed Sora to generate videos featuring characters from Disney, Marvel, Pixar, and Star Wars, a move initially seen as a landmark for the AI industry. However, with the decision to shut down Sora, this deal has also collapsed, though reports indicate that no money had actually changed hands before its termination. Disney issued a statement respecting OpenAI’s decision and affirming its continued engagement with AI platforms to explore new ways of connecting with fans while respecting intellectual property rights.

Despite an initial surge in interest, Sora failed to maintain sustained user engagement. The app peaked in downloads in November, reaching approximately 3.3 million across iOS and Android platforms. However, this momentum proved short-lived, with downloads plummeting to about 1.1 million by February. In comparison to ChatGPT’s 900 million weekly active users, Sora’s performance indicated a lack of lasting appeal. The app reportedly generated around $2.1 million from in-app purchases, which allowed users to acquire more video generation credits. While not a significant revenue stream, the app’s considerable computing demands and the mounting legal and ethical liabilities likely contributed to OpenAI’s decision to discontinue it, especially given its lack of sustained growth.

The shutdown of the Sora app, however, does not signal the end of the technology itself. The powerful underlying Sora 2 video- and audio-generation model remains available, albeit integrated behind the ChatGPT paywall. Experts warn that the threat of easily accessible AI deepfake technology persists, and it is merely a matter of time before new social AI video applications emerge, potentially leading to another wave of artificially generated content, highlighting the ongoing challenge for moderation and responsible AI deployment.