OpenAI Launches AI Video Generator Sora in the US Amidst Criticism and Optimism

OpenAI Launches AI Video Generator Sora in the US Amidst Criticism and Optimism

Image credit: OpenAI

OpenAI is announcing Sora, an AI video maker, as a publicly available tool in the US. Sora, a so-called “text-to-video generator,” allows users to create short video clips on the basis of given texts. This new tool was developed in February and then shown exclusively to a small number of artists, filmmakers, and safety testers. OpenAI’s Sora release represents its ambition of redefining video storytelling with new dimensions of innovation and creativity.

An example OpenAI highlighted: a serene video of woolly mammoths trekking through sand dunes, generated from the very simple text prompt, “a wide, serene shot of a family of woolly mammoths in an open desert.” OpenAI hopes, in a blog post, that Sora will inspire creators all over the world to extend storytelling boundaries.

And notwithstanding the fanfare, the product has not had an altogether smooth ride since its introduction. On its launch date, OpenAI’s website reported surges in traffic with new sign-ups temporarily restricted. The tech has its very vocal critics as well: one tech reviewer, Marques Brownlee, described the first generation of this technology, also known as Sora for “sort of like reality,” “horrifying and inspiring”-great at creating landscapes, stylistic effects, but terrible at mimicking even the most basic physics.

Other filmmakers who had early access to Sora also said the tool was buggy and had odd visual artifacts and other defects. OpenAI copped to those limitations, saying that the tool was still evolving. The company has also sought to reassure users of their commitment to the improvement of Sora’s functionality.

Notably, its international rollout has been restricted because of regulatory challenges. The tool is not available in the UK and European Union as OpenAI works its way through the Online Safety Act, Digital Services Act, and GDPR. Two weeks ago, OpenAI cut off access to Sora after a group of artists dubbed the “Sora PR Puppets,” discovered a backdoor to game the tool. The artists called out OpenAI for using them to shill the tool under false pretenses while undermining their livelihoods.

Critics of generative AI have raised concerns about its misuse, including its potential to spread misinformation and create deepfakes. Examples of deepfakes featuring prominent figures like Ukrainian President Volodymyr Zelenskyy and US Vice President Kamala Harris have sparked fears of political disinformation campaigns. OpenAI has implemented safeguards to limit such abuses, including banning uploads involving specific individuals and blocking explicit content.

Currently, Sora is available to paying subscribers of OpenAI’s services in the US and several other countries but remains inaccessible in Europe. The ongoing effort by OpenAI to address compliance and technical challenges makes Sora the embodiment of both the promise and peril that generative AI technology holds for creative industries.