Fact Checked by 
OpenAI has made its Sora AI video generator available to the public, including some tools that help creatives produce complex projects.
The long-in-development model lets you generate short video clips from text prompts. However, you can also “remix” prompts to varying degrees by using additional prompts, and a Storyboard can string together multiple prompts while producing transitions between clips.
Sora limits videos to resolutions between 480p and 1080p for up to 20 seconds, and OpenAI has taken numerous steps to minimize the potentials for copyright violations and misinformation that affect both the company as well as rivals like xAI’s Grok. You can’t base videos around many forms of copyrighted material or some recognizable public figures, and all footage includes a watermark.
You need a ChatGPT Plus subscription to create up to 50 videos at 480p, and fewer at 720p. Pro users get 1080p access, 10 times the usage, and longer-lasting videos. OpenAI says it’s working on “tailored pricing” for a variety of users starting early 2025. Free ChatGPT users can always watch clips, but can’t create them.
OpenAI Sora: potential and pitfalls for video production
As with other AI video generators, Sora’s results are frequently mixed. You might see continuity errors like disappearing objects or misspelled text. The large language model also doesn’t have a built-in understanding of physics, so objects and effects might not behave properly. This isn’t something you’d use to replace conventional video production, especially for realistic output.
However, it might be useful for animations and abstract videos where accuracy isn’t important. You could make persuasive motion graphics or B-roll footage without requiring significant artistic knowledge.
There are some concerns that might persist even if OpenAI addresses all the technical issues. Models like Sora and Google’s Veo still tend to be trained on publicly available content, raising the possibility that they’ll violate copyright. Actors and producers have already objected to the potential that their work might be replaced by AI — Sora won’t completely alleviate their fears.
And while OpenAI might make it harder to produce misinformation, you can still crop watermarks (as YouTuber Marques Brownlee illustrated) and use carefully-written prompts to mislead audiences. There’s still the chance that malicious creators will use Sora and similar generators to spread false claims that could lead to real-world harm.
