Meta will begin annotating AI-generated photos from OpenAI and Google

Meta will begin annotating AI-generated photos from OpenAI and Google

Meta will begin annotating AI-generated photos from OpenAI and Google

In an attempt to inform users that the images—which often resemble genuine photos—are really computer creations, Meta will add the label to any material featuring the identifier shared on its Facebook, Instagram, and Threads sites, the firm said. The company's head of global affairs, Nick Clegg, said in a blog post.

Meta will begin annotating AI-generated photos from OpenAI and Google

In the next months, the Meta platform will start using a set of invisible markers integrated into files to identify and categorize photos produced by other firms' artificial intelligence services, the company's chief policy officer said on Tuesday.

In an attempt to inform users that the images—which often resemble genuine photos—are really computer creations, Meta will add the label to any material featuring the identifier shared on its Facebook, Instagram, and Threads sites, the firm said. The company's head of global affairs, Nick Clegg, said in a blog post.

Any material produced by the corporation using its own AI techniques already has labels.

According to Clegg, once the new method is operational, Meta will apply the same to photos generated on platforms hosted by OpenAI, Microsoft, Adobe, MidJourney, Shutterstock, and Google, an Alphabet company.

The statement offers a sneak peek at a new set of guidelines being worked on by tech firms to lessen the possible negative effects of generative AI systems, which produce phony but realistic-looking material in response to basic commands. Can't.

This strategy expands upon a model that many of the same corporations have developed over the last ten years to manage the removal of prohibited information from various platforms, such as images of child abuse and widespread violence.

In a Reuters interview, Clegg expressed his confidence that businesses could already accurately tag AI-generated photos, but he acknowledged that the technologies needed to identify audio and video information were still being developed and more complicated. are being carried out.

Clegg said, "Despite the fact that the technology isn't fully mature yet, particularly whenever it comes to audio and video, hopefully we can create an impression of momentum as well as encouragement for the rest of the industry."

He said that Meta would start enforcing sanctions for anyone who fail to mark their own changed audio and video material in the meantime. Clegg did not elaborate on the penalty.

According to him, there isn't a workable method for labeling textual content produced by ChatGPT or other AI tools at the moment.

"That vessel has departed," Clegg said.

If generative AI material is posted on WhatsApp, the business will designate it as such, a Meta representative refused to comment.

The company's policy on doctored films was criticized by Meta's independent review board on Monday, who said that it is deceptive and that material should be tagged rather than erased.

Clegg said that he generally agreed with such critiques.

He said that the board was correct to conclude that Meta's current policy "is not fit for purpose in an environment where you are going to have far more synthetic materials as well as hybrid materials than before."

He pointed out that Meta is already heading in the path the board had suggested with the establishment of the new labeling partnership.