OpenAI launches Sora: the ChatGPT AI video generator is now widely available
Now everyone can create fake videos, with limitations, of course
đ OpenAI officially launched its controversial AI Video generator Sora
đš ChatGPT Plus and Pro subscribers can generate up to 500 priority videos
đ ââď¸ This version of Sora still has âmany limitationsâ as demand has soared
đ All Sora-generated videos come with embedded metadata to verify origin
đŤ Illegal, damaging and abusive content is blocked along with deepfakes
đââď¸ Didnât make the cut: uploads of people are limited at launch
If AI wasnât controversial enough already, OpenAI just launched its video-generating software called Sora. The company previewed this technology earlier this year with a series of videos that stunned quite a few people with just how realistic the videos looked. The launch is part of OpenAIâs â12 Days of Ship-masâ program for December and is available in the US and âmost other countriesâ where ChatGPT is already available.
When Sora was previewed back in February, there was a lot of hype around the 90-second videos that showcased a range of settings, humans, and actions. At launch, OpenAI is limiting Sora content to 1080p videos at 20 seconds long. The AI company was also candid in its blog post about the limitations of this new technology: âIt often generates unrealistic physics and struggles with complex actions over long durations.â This likely contributed to the 20-second length cutoff.
OpenAI also developed a new interface for Sora to make it easier to interact with. Soraâs new storyboard tool gives users control over each frame, letting them specify which input theyâd like for each moment of the video. Users will also be able to view Featured and Recent content created from members of the community (free account holders can view these as well).
Questions about Sora remain
Obviously, one of the biggest questions surrounding a tool like this is safety and protecting people from malicious, illegal, and inappropriate content. To that extent, OpenAI specifically mentioned built-in tools within Sora to block child sexual abuse materials (CSAM) as well as sexual deepfakes, self-harm, violence, deceptive election content, and more.
Similarly, OpenAI says itâs limiting uploads of real people to a small group of testers at launch, but it wouldnât be too hard to create a fake video with someone who looks oddly similar to anyone with a lot of videos and images of them already out there.
Another huge question regarding a tool like this is the profound impact it will have on the entertainment, marketing, and advertising industries. As someone who has worked in two out of three of them, I find it a bit unnerving to see what Sora could be capable of doing in just a few short months and years.
Honestly, I canât see any marketing agency - especially smaller ones with more limited budgets - hiring production studios for thousands of dollars to make commercials when it can get Sora to do it for $200 a month. Just as ChatGPT could quite easily come for the journalism industry very soon (donât worry, itâs still a human typing this), Sora could just as easily wipe out thousands of jobs in just a few short years.
Clearly, there are still many, many questions to be answered about Sora, and OpenAI was very light on the details at launch. And while it
But donât worry, itâs still going to change the world, right? At least thatâs what the billionaires are telling us.
Jason Cockerham is a seasoned technology journalist reporting on the latest tech trends for The Shortcut as well as CNET, Android Police, XDA Developers and Top Ten Reviews. You can follow him at @jasonthejasonc on X and LinkedIn.