AI-Generated Video App Tops App Store, Raises Eyebrows
OpenAI’s newly launched Sora 2 app is dominating social media just one day after release, topping the iOS App Store’s Photo and Video category. The app allows users to generate lifelike videos from text prompts, embedding themselves or others into surreal, AI-generated clips — from Sam Altman singing in a toilet to James Bond-style poker scenes.
Powered by the upgraded Sora 2 media model, the app creates high-definition videos using text inputs and audio-visual scans of real people. With viral potential baked into its design, Sora 2 also introduces a host of copyright and misinformation concerns. Users have already used copyrighted characters like Mario, Pikachu, and Ronald McDonald in their videos, raising red flags among legal experts.
Copyright Clash and Deepfake Dangers
Critics argue that Sora’s approach to copyright law is dangerously aggressive. According to reports, copyrighted material can appear in Sora videos unless rightsholders opt out — and not through a blanket request, but by submitting specific examples. Legal scholars warn this method runs counter to how copyright protection actually works.
OpenAI is already facing lawsuits from authors and publishers, including The New York Times, for alleged copyright violations in training its AI models. Meanwhile, competitors like Anthropic have paid hefty settlements to resolve similar claims.
Sora’s capabilities also reignite concerns about deepfakes. One viral video showed OpenAI CEO Sam Altman stealing computer parts from Target — entirely fictional but highly realistic. Experts fear such content could be used to depict people in crimes or behaviors they never committed.
Transparency Tools and Their Limitations
In response to misinformation risks, OpenAI has implemented safeguards like visible moving watermarks and invisible metadata tagging to signal that Sora 2 content is AI-generated. However, these measures are far from foolproof. The metadata can be stripped away when videos are uploaded to social platforms, and experts say watermark effectiveness remains unproven.
Siwei Lyu, a computer science professor and media forensics expert, noted that these tools require more testing. “Their effectiveness can only be judged internally at this point,” he said. OpenAI has acknowledged the limitations in its own safety documentation and pledged continued improvements.
Access Still Limited, But Demand Surging
Despite being available for download, the Sora app remains invite-only for actual use. OpenAI is gradually expanding access while monitoring legal, ethical, and technical responses. The early hype has made Sora 2 the centerpiece of the AI-generated video space — and a likely flashpoint in ongoing copyright and misinformation debates.
As video generation tools gain realism and accessibility, the question now facing regulators and tech companies alike is how to balance innovation with responsibility.
