Meta fails to curb the spread of many sexual AI Deepfake celebrity images on Facebook

Meta (Meta) in CBS News Investigation Deep bubble images processed by AI On the company's Facebook platform.

Actors Miranda Cosgrove, Jeanette McCurdy, Ariana Grande, Scarlett Johansson and former tennis star Maria Sharapova actors Miranda Cosgrove, Jeanette McCurdy, Ariana Gran Dozens of fake, highly personalized images by Ariana Grande and former tennis star Maria Sharapova have been shared by multiple Facebook accounts and have gained hundreds of thousands of likes and many Similar scenes.

“We have removed these images to violate policy and will continue to monitor other violations of posts. This is a challenge across the industry and we are constantly working to improve our detection and law enforcement technologies.” In a statement sent by email on Friday.

The analysis of these images by the reality defender is a platform that detects AI-generated media, showing that many of the photos are Deepfake images – AI-generated, with corpses of underwear replacing the body of celebrities, to replace the body of celebrities, and otherwise replacing the real photos of celebrities. According to Reality Defender's analysis, some images may have been created using an image stitching tool that does not involve AI.

Ben Colman, co-founder and CEO of Reality Defense Attorney, told CBS News on Sunday. “The growth rate of such content is growing, especially when there are few existing measures for such content.”

CBS News reports on Miranda Cosgrove, Jeanette McCurdy, Ariana Grande and Maria Sharapova Sharapova) comments. The actor's representative said Johnson refused to comment.

Under Meta's bullying and harassment policies, the company prohibits “derogatory Photoshop or drawings” on its platform. The company also bans adult nudity, sexual activity and adult sexual exploitation, and its regulations aim to prevent users from sharing or threatening to share involuntary sexual intimacy images. Meta has also launched the use of the “AI Information” tag to clearly mark the content of AI manipulation.

However, there are still questions about the effectiveness of tech companies policing such content. CBS News found that dozens of AI-generated, personalized Cosgrave and McCurdy were still publicly available on Facebook and were tagged as Meta even after widespread sharing of such content in violation of company terms.

There is still a deep image of Cosgrave that is still released on weekends, which is shared by 2.8 million followers.

Both actors are Icarly of Paramount Global, a parent company of CBS News, both actors are former child stars of Icarly of Paramount Global, a parent company of CBS News, according to public figures analyzed by CBS News.

Meta's Oversight Committee is a quasi-independent body composed of experts in the field of human rights and freedom of expression and makes recommendations for content review on the Meta platform. insufficient.

The Oversight Committee cites advice on the dollar over the past year, including urging the company to clearer rules by updating its ban on “derogatory Photoshop” to include specifically the term “non-consent” and include other photo operations such as AI Techniques like that.

The board also recommended that it prohibit “derogatory Photoshop” from inclusion in the company's adult sexual exploitation regulations, so the review of such content will be implemented more strictly.

When asked about the board’s recommendations on Monday, Meta pointed to guidelines on its transparency website that the company has rejected the recommendations so far, although Meta noted in a statement that it is still considering how it shows a lack of consent The AI-generated image. Mehta also said it is considering reforms to its adult sexual exploitation policy to “capture the spirit of the board’s recommendations.”

“The Oversight Committee has made it clear that involuntary deep private images are serious violations of privacy and personal dignity, and the damage to women and girls is disproportionate. These images are not only an abuse of technology – they are a kind of abuse Form, can last forever,” Michael McConnell, co-chair of the Oversight Committee, told CBS News on Friday.

“The board is actively monitoring Met’s response and will continue to push for stronger safeguards, faster law enforcement and greater accountability,” McConnell said.

Meta isn't the only social media company that has to face widespread, sexually-inspired content issues.

Last year, Elon Musk's platform X temporarily blocked Taylor's quick related searches, after AI-generated fake images circulated widely on the platform as singers and gained millions of Perspectives and impressions.

“Only banned involuntary nude (NCN) images on X, we have a zero tolerance policy for such content,” the platform's security team said in a post at the time.

A study released by the UK government earlier this month found that the number of DeepFake images on social media platforms has expanded rapidly, with the government expecting to share 8 million deep hits this year, from 500,000 in 2023.