Users Have One Big Problem with Sora
Briefly

Users Have One Big Problem with Sora
"OpenAI's Sora, the TikTok-like video generator app, was an instant hit, shooting to the top of the download charts and hitting a million installs within a few days of launch. Its early popularity makes sense: It's a free chance to try recent video generation technology, a fun way to joke around with friends who consent to being included in one another's clips, and an early look at the weird dynamics of celebrity-endorsed AI."
"Judging by App Store reviews, though, it seems like people aren't entirely happy with Sora. Some of the complaints dragging on its rating - it's at 2.8 stars out of 5, compared to 4.9 for ChatGPT or 4.6 for Google - are straightforward and easily addressed: it's sort of glitchy; the voices aren't as good as they could be; some generations are uncanny or unconvincing; the app, despite being free, is invite-only, which can make for a disappointing download."
"Running into Sora's "guardrails," as the company explicitly calls them, is a frequent experience if you're using the app to, say, goof off with your friends. (The app's public feed provides plenty of evidence that people are using this app to tease and embarrass their friends and the few public figures who have given permission for their likenesses to be used, like Sam Altman.)"
Sora is a TikTok-like video generator that achieved one million installs within days by offering free, emergent video-generation features and playful social use cases. Early users treat the app as a way to joke with consenting friends and to experiment with celebrity-endorsed AI. App Store reviews reveal notable dissatisfaction: a 2.8-star rating reflecting glitches, weaker voices, uncanny or unconvincing outputs, and an invite-only barrier. The most frequent complaint concerns extensive content guardrails that block common prank-oriented usages. OpenAI has implemented many limits to prevent harms associated with deepfakes, creating tension between safety and user expectations.
Read at Intelligencer
Unable to calculate read time
[
|
]