Identifying AI-Generated Content with the Trust Label 'Made with AI' | Release #106
Image by らっしぃ
As AI rapidly evolves, its convenience permeates our daily lives. However, we often question whether content created by AI is a real photo or a generated image.
In this context, the introduction of the 'Made with AI' label by Meta seems to be a step towards alleviating our concerns and deepening our understanding of AI content.
Introduction of the 'Made with AI' Label
With the advancement of AI technology, content generated by AI is increasing. Meta has announced that starting next month, videos, images, and audio generated by AI will be labeled 'Made with AI'.
This initiative applies to platforms under Meta, such as Facebook, Instagram, and Threads, aiming to clarify the origin of content for users.
Determining Content Authenticity through C2PA
The movement towards enhancing reliability and transparency is not limited to Meta. Companies like Microsoft, Intel, Adobe, ARM, BBC, and Truepic have established C2PA to record the origin and authenticity of digital content. The technical specifications formulated by this coalition serve as a foundation for ensuring content reliability.
YouTube and OpenAI's DALL-E require creators to disclose when posting AI-generated content. Particularly with DALL-E, since February this year, generated images include C2PA metadata, making it easier for consumers to understand how the content was created.
Different AI Approaches by Platforms
Efforts to enhance the reliability of AI content are still in the early stages. There are numerous challenges to address, such as handling unlabeled content and improving algorithms. Addressing these challenges can further improve the management and control of AI content.
On platform X, the official stance on AI content has not yet been revealed. Depending on the platform's policy, there could be waves of user migration. As the digital era progresses, platform responses will significantly impact content reliability.
