Calvin Wankhede / Android Authority
I recently compared ChatGPT’s Sora 2 against Google’s Veo 3 AI video generator and came away impressed by how far OpenAI had pushed the boundaries of the technology with its realism and accurate physics. At the time, Sora 2 was only accessible via the web or a dedicated iOS app. Now, OpenAI has brought the Sora app to Android and it’s more than just a basic AI video generator — it’s a full social media platform similar to TikTok or YouTube Shorts that you can also personally star in.
And while platforms like Midjourney have allowed you to share your AI creations with others, the new Sora app takes things one step further — it allows you to add your own face and voice to the video. So how does Sora achieve this bizarre feat and what’s it like to use? I installed it to find out.
Sora is more than an AI video generator
The Sora app feels familiar when you open it; the main feed scrolls vertically like any short-form video platform, and even includes the like, comment, and share buttons where you’d expect to see them. This feed is dominated by AI-generated creations, of course, but it’s not all artificial slop. As I mentioned earlier, Sora allows you to put yourself in an AI-generated video. Scroll through the feed and you’ll come across the likeness of various personalities, ranging from OpenAI CEO Sam Altman himself to famous YouTubers.
This is all made possible through Sora’s headlining Cameo feature, which scans your facial features using just your phone’s selfie camera. What I found most surprising was that the scanning process took less than ten seconds, faster than setting up face unlock on my Pixel. Once complete, I could immediately begin generating “cameos” or AI-generated videos with my face and voice. To achieve this, I clicked create and included my username and a description of the video.
This worked surprisingly well — my body’s proportions are slightly off in the AI-generated videos but the skin tone and other features line up near perfectly. It feels uncanny to see yourself in scenes that never happened, and my friends remarked they wouldn’t have known the videos were AI generated if it wasn’t for the watermark.
Sora can place you in any environment, no matter how absurd, and will even replicate your voice with any text you include in your prompt. If you know other people on the platform, you can share your avatar with them. They can then use your likeness in videos of their own. Or you can open up your likeness to the world, as some influencers have done. Historical figures can also be included in your creations with few exceptions. Here are a couple of examples I generated where Bob Ross and Sam Altman explain app sideloading:
But while many of Sora’s features focus on marrying AI with real humans, the app’s feed feels like the exact opposite. You’ll find no shortage of cat videos, but there’s little else worth your time. Still, I can’t deny there’s some appeal in being able to place yourself in unique and interesting settings without publicly sharing the videos.
On iOS, the Sora app also allows you to upload videos of objects and animals and add them to your cameo. This means you could have your pet appear alongside you in an AI generated video. This feature hasn’t made its way to the Android app yet, though.
Would you use an AI video generator if you could add your likeness?
191 votes
Is Sora worth downloading?
Ryan Haines / Android Authority
Sora 2 was invite-only at launch, but it’s now open to anyone living in the US, Canada, South Korea, and Japan. If you live outside of these countries, you’ll still need an invite code from an existing user or wait until the platform finishes rolling out.
Once you gain access to the Sora app, though, OpenAI will let you generate 30 videos per day. The platform is also completely free, and doesn’t include any advertisements. By contrast, Google requires you to be a paying AI Pro subscriber to use Veo 3 and you can only generate a handful of eight-second videos per day. OpenAI is clearly absorbing Sora’s monumental compute cost for now.
As for whether you should use the Sora app, I’m conflicted. On one hand, it’s a shocking perspective on how far generative AI has come that it can mimic content you’d expect to see a human create on Instagram or YouTube. In the early days, Sora’s guardrails were practically non-existent. You could generate copyrighted characters, real public figures, even historical icons. I saw a clip where Martin Luther King Jr. yelled profanities into crowds and a recreation of Titanic with Pokemon instead of human actors.
Sora has incredible breadth, but nearly zero depth.
Needless to say, OpenAI was forced to clamp down after widespread complaints from copyright holders. So what’s left on the Sora app now? Admittedly, not a whole lot.
Sora might be useful as a form of surreal escapism for those that want to tune out the world around them, but the AI model’s hyper-realism doesn’t exactly lean towards this aspect. Even if you use Sora as OpenAI intends: a way to place yourself in a variety of settings that you can’t recreate in the real wold, I’m left wondering who the audience is.
After an hour of scrolling through Sora’s seemingly endless feed of AI-generated clips, I felt nothing. There’s no narrative thread, no humanity, no context to any of the videos on Sora. Unlike Instagram or YouTube Shorts, where I might stumble upon some educational content or a slice of someone’s life, Sora is completely synthetic. It’s entertainment for entertainment’s sake that’s utterly detached from the real world.
Ultimately then, I suspect most people will use the Sora app as an AI video generator and ignore the social aspects altogether. And if you’re willing to hand over a scan of your face to OpenAI, I’d recommend trying it out at least. The tech aspect of it is extremely impressive, and there’s nothing else like it.
Thank you for being part of our community. Read our Comment Policy before posting.
