Meta announces labeling of AI-generated images on Instagram, Facebook, and Threads

On February 6, 2024 local time, Meta announced that it plans to detect and label AI-generated images posted on Instagram, Facebook, and Threads. This feature will be rolled out to all languages in the coming months at the earliest.

Labeling AI-Generated Images on Facebook, Instagram and Threads | Meta

Meta to start labeling AI-generated images from companies like OpenAI, Google | Reuters

Meta pushes to label all AI images on Instagram and Facebook in crackdown on deceptive content | Meta | The Guardian

According to Meta, detection and labeling of AI-generated images is done by using invisible metadata embedded in the image file. Specifically, we are working with companies such as Google, OpenAI, Microsoft, Adobe, Midjourney, and Shutterstock to develop industry-standard metrics to easily identify AI-generated information .

Therefore, until now it was only possible to label images generated by Meta's proprietary image generation model ' Emu ,' but in the future it will be possible to label images generated by image generation models from various companies. becomes possible.

Posts containing AI-generated images will be labeled “AI info,” making it easier for users to identify the images as AI-generated and increasing transparency. Please note that posts containing images generated using Emu are already labeled as 'Imagined with AI.'

Nick Clegg, president of global affairs at Meta, said: 'As the line between human-generated content and AI-generated content blurs, people want to know where the line lies. ” states.

Clegg said the feature is in the construction phase and will begin applying labels in all languages supported by each app in the coming months. Mr. Clegg wrote that he hopes to implement this feature by 2025, when several important elections will be held around the world.

Meta has also revealed that it plans to apply this feature to content including audio and video in the future. 'Going forward, we will require users to label AI-generated content and impose penalties if they fail to do so,' Clegg said. We will place more prominent labels on AI-generated content that poses a higher risk.'

'People and organizations that actively seek to deceive people with AI-generated content will look for ways to circumvent the safeguards in place to detect AI-generated content,' said Clegg. In our industry and society at large, we must continue to look for ways to stay ahead of the curve.'

in Software,   Web Service, Posted by log1r_ut