ИИ-фейки не пройдут: сервис для маркировки созданного с помощью нейросетей контента создают в России

Микроскопические маркеры в фото и видео помогут бороться с фейками

The Institute for System Programming of the Russian Academy of Sciences and Yandex are developing a technology for marking images, videos, and audio created using artificial intelligence. The new system will be invisible to humans - a microscopic digital marker is embedded in the content, which can be read by special analyzers.

The service will allow checking whether the material was generated by a neural network, in particular, using "Shedevrum" - an image generator from Yandex. The developers are also considering making the technology open to the entire market.

The development of generative AI creates risks of spreading fakes and disinformation. Hidden marking technology will help distinguish real content from artificial content, increase transparency and trust in digital information.

Read more on the topic:

Created by a non-human: the State Duma proposed to mark content created by neural networks

Technology development or security?: The State Duma is discussing a bill "On regulation of artificial intelligence systems in Russia"

Moscow is developing a technology for marking AI-generated content