AI for automatic content filtering created in Novosibirsk

The system masks obscene language and 18+ materials

The press service of Novosibirsk State University (NSU) has developed an artificial intelligence (AI) system that automatically hides unwanted fragments in video and audio recordings. The solution is intended for users, content authors, media companies and video platforms.

The system analyzes materials for unwanted content. It covers detected fragments with a mosaic or replaces the sound with a “beep”. According to the developers, this allows publishing content that complies with platform rules and legislation without manual moderation.

The developers studied the market and stated that there are no analogues to their product. Most existing solutions work only with images, require moderators, or only notify about violations. The new system automatically hides problematic elements, which simplifies the publishing process and reduces risks for platforms.

The service is especially relevant for media platforms that are responsible for user content. It automatically adapts materials to legal requirements, which minimizes fines and blocking, the press service noted.

The team is currently completing testing and integration of the system. In the near future, it is planned to be released in two versions: a free one for private users and a commercial one for large companies.

 

Spread the love