Facebook And Instagram To Start Labelling AI-Generated Content
You can now identify AI content, including images, videos, and audio.
By Tharshini Skovalan - 08 Feb 2024, 05:33 AM
Cover image via Meta can now identify AI-created images on Facebook, Instagram, and Threads to ensure they are recognised as artificial
This initiative is aimed at combatting misinformation involving the use of generative AI. The company confirmed recently that it will be implemented in the coming months.
Image via Meta
Meta CEO Mark Zuckerberg plans to develop general artificial intelligence using data from Facebook and Instagram
The company's decision to label AI-generated content follows a ruling by their Oversight Board about a misleading video, which necessitated measures to improve community guidelines.
Meta is teaming up with industry partners to establish standards for AI-generated content
The company is now planning to gather insights, transparency preferences, and new technology. This will help them improve their policies as they expand these features.
Meta is working on tools to detect hidden signals in accordance with C2PA and IPTC standards. For example, images from places like Google, OpenAI, Microsoft, Adobe, Midjourney, and Shutterstock can be easily identified and labeled as AI-generated.
However, the ability to detect AI-generated video and audio is not as advanced as that of images.
Image via Meta
Meta wants to prevent people from deceiving others
The company will penalise users who don't reveal AI-generated content. Users must use watermarks and automatic detection tools developed by Meta's FAIR AI research lab.
Additionally, any photorealistic images created with Meta's AI tool will include a label saying 'Imagined with AI'.
