Using beauty-filtered photos as advertisements is illegal! Label them as "photo fraud", Norway takes action against celebrities and internet celebrities

Xiaoxiao sent from Aofei Temple
Quantum Bit Report | Public Account QbitAI

Are you still worried that the photos of internet celebrities you see are "fake" and don't look like the real person at all?

Now, some countries have enacted laws to explicitly prohibit the use of unlabeled advertising photoshopped images!

In other words, if celebrities and internet celebrities want to do paid promotion, they can only use the most original photos; if they must post photoshopped photos, they must also indicate that they are "photo frauds", otherwise they will be fined .

The law was passed overwhelmingly with 72 votes to 15, with strong support.

Let's take a look at what this bill is.

Targeting celebrities and internet celebrities to crack down on "photo-scam" advertisements

In fact, this is the amendment bill made by Norway in June this year regarding "modified advertising labels" in the Marketing Act.

The revised content is as follows:

Advertisers and influencers will have to label ads that show bodies being altered in shape, size or skin, whether using Photoshop or other manipulation.

Judging from the content, whether it is to change the body shape by slimming the waist and face , to adjust the body size by plumping the lips and increasing muscle , or to change the skin by whitening and removing acne , they are all labeled as "photo fraud".

From the perspective of the target, this bill is not aimed at everyone, but only at "influential people" such as Internet celebrities and celebrities , as well as advertisers and other groups with commercial interests.

From the perspective of platforms, this bill applies to all mainstream social media platforms abroad, including Instagram, Facebook, Snapchat, and TikTok.

The bill was proposed by Norway's Ministry of Children and Family Affairs.

More than half of year 10 girls in schools in the Norwegian capital Oslo struggle with mental health, department research shows, with anorexia the third most common cause of death among teenage girls.

This type of advertisement and media will put tremendous pressure on young people, making them pursue "impossible" beauty.

It sounds reasonable and can prevent advertisers from using "photo fraud" to seek improper benefits, while also protecting the physical and mental health of young people.

But it is still difficult to actually implement it.

How to determine whether a picture has been photoshopped?

Although the bill was passed, how to enforce it was a bigger problem.

For example, how to enforce it for groups living outside Norway and not subject to Norwegian law? Does adjusting the brightness and contrast of a photo for the sake of overall perception fall under the category of photo retouching?

In addition, there is currently no relevant department or technical staff to determine whether a photo has been photoshopped.

According to Vice, the law could also have side effects, such as encouraging influencers such as internet celebrities to opt for plastic surgery to maintain their photo-realistic looks.

It is not yet known how Norway will check photoshopped photos. However, the AI ​​community already has many algorithms that specifically combat Deepfakes, photoshopped photos, and other images and videos that have been modified by AI.

The "cat and mouse" game between the two sides has never stopped.

For example, Adobe, the "ancestor of photo retouching", invented a set of "reverse photo retouching" tools as early as 2019 .

This tool can not only identify whether a photo has been retouched, but also restore the retouched photos using AI:

Simply put, it is to feed the AI ​​with photos that have been photoshopped and those that have not been photoshopped at the same time, so that the AI ​​can learn to recognize the difference between the two photos and actively generate an "optical flow field" on the photoshopped photos.

The restoration effect is also good, very close to the original image:

Facebook, together with Michigan State University, has also launched a Deepfake "reverse engineering" that can use the features left by the model in images and videos to conduct reverse analysis to find the source of fake data.

The projects involved include these fake faces generated using GAN:

This kind of "cat and mouse" game of AI identifying AI will continue, and may also serve as one of the examples for the official to identify P-pictures in the future.

what you think?

Norwegian regulations portal:
https://www.stortinget.no/no/Saker-og-publikasjoner/Vedtak/Beslutninger/Lovvedtak/2020-2021/vedtak-202021-146/

Reference links:
[1]https://m.weibo.cn/status/4656195858595946
[2]https://petapixel.com/2021/07/02/not-disclosing-that-a-photo-was-retouched-is-now-illegal-in-norway/
[3]https://www.dailymail.co.uk/news/article-9748731/Norway-makes-ILLEGAL-influencers-post-retouched-photos-without-saying-theyve-edited.html
[4]https://zhuanlan.zhihu.com/p/69840591
[5]https://www.theverge.com/2021/6/16/22534690/facebook-deepfake-detection-reverse-engineer-ai-model-hyperparameters