[ad_1]
OpenAI launched its video creation tool, Sora, on Monday. But the company has chosen not to release a key feature to most users pending further testing.
The feature in question creates a video using an uploaded photo or footage of a real person as a reference. OpenAI says it will give a “subset” of Sora users access to it, but won’t roll out the capability broadly until it has had a chance to fine-tune its “approach to safety.”
“The ability to create a video using an uploaded photo or video of a real person as the ‘seed’ is a vector for potential abuse, and we are taking a particularly incremental approach toward learning from early usage patterns,” OpenAI wrote in a post. Blog post. “Early feedback from artists suggests that this is a powerful creative tool that they value, but due to the potential for abuse, we are not initially making it available to all users.”
OpenAI also Users will not be allowed to share created videos that contain clips or photos of real people to Sora’s homepage discovery feed.
“We obviously have a big target on our back as OpenAI, so we want to prevent illegal activity with Sora, but we also want to balance that with creative expression,” Rohan Sahai, Sora’s product lead, said during a live presentation earlier today. “We know (this is going to be) an ongoing challenge — we may not get it perfect on day one. We’re starting a little conservative, so if our moderation isn’t right, just give us that feedback.
Generative video is a powerful tool — and a controversial one, with deepfakes and misinformation a major concern. According to According to data from identity verification service Sumsub, deep fraud worldwide has increased more than 10 times from 2022 to 2023.
Among other steps OpenAI says it’s taking to prevent abuse, Sora has a filter to detect if a generated video depicts someone under 18. If so, OpenAI applies a “tougher threshold” for moderation regarding sex, violence, or self. -Harmful content, as the company claims.
All videos created with Sora contain metadata to show where they come from – specifically metadata that adheres to C2PA technical standards. Metadata can be removed and granted. But OpenAI is touting it as a way for C2PA-enabled platforms to quickly detect whether a video originated from Sora.
In an effort to stave off copyright complaints, OpenAI also says it is using “hot rewrites” to prevent Sora from creating videos in the style of the live creator.
“We’ve added quick rewrites designed to trigger when a user attempts to create a video in the style of a live artist,” the company wrote. “We’ve chosen to take a conservative approach with this release of Sora as we learn more about how the creative community uses Sora… There’s a very long tradition of creativity drawing on other artists’ styles, but we appreciate that some creators may have concerns.”
A number of artists have sued AI companies, including OpenAI, for allegedly coaching their work without permission to create AI tools that recreate content in their unique styles. For their part, the companies have claimed that the principle of fair use protects them from copyright infringement claims, and that AI models are not replicated in reality.
In terms of training, OpenAI will only say that Sora was developed using a mix of “diverse data sets,” including publicly available data, private data accessed through its partnerships with data vendors, and custom sets developed internally. Earlier this year, former OpenAI CTO Mira Moratti explicitly denied that Sora had apparently been trained on YouTube clips. violation Google-owned streaming platform usage policy.
According to vlogger Marques Brownlee, who got an early preview of Sora, the system can create multiple variations of videos from a text or image prompt and edit existing videos via a remixer. The Storyboard interface allows users to create sequences of videos; The Blend tool takes two video clips and creates a new video that preserves elements of both; Loop and Re-cut options allow creators to further edit and tweak their videos and scenes.
Subscribers to OpenAI’s ChatGPT Pro and Plus plans get access to Sora, OpenAI’s video creator, today — but only if they live in certain countries.
[ad_2]