Beauty filters are now powered by AI, should we be scared?

Technology is helping developers churn out increasingly realistic face alteration tools that are becoming popular on social media

A before-and-after look of TikTok's Bold Glamour effect. Photo: joannajkenny / TikTok

Sharp contours, symmetrical eyebrows, smoother and brighter skin – these are just some of the facial alterations offered by the TikTok effect Bold Glamour.

Beauty filters on social media are not new, but this one, which became available in February, is particularly remarkable. It barely distorts or glitches, unlike other effects, creating an eerily seamless “transformation”.

The filter has now been used more than 75 million times and, while impressed, TikTok users were quick to warn about its potential impact on people's self-esteem.

Body and skin-positivity creator Joanna Kenny recently demonstrated how natural the filter looked, in a video that has been viewed nearly 12 million times.

“It's crazy,” she says. “I don't actually look like this. I don't want to say this about myself, but I actually feel immediately ugly when I take this filter off.”

In the caption, she warns TikTok users to avoid using the filter.

“Filtered skin is not a skin type,” she adds.

Many other creators on the platform have echoed Kenny's sentiments, with comments alluding to how the filter strongly reinforces unrealistic beauty standards.

The results are often “unobtainable, false or entirely impossible to achieve” in real life, integrative psychotherapist Sol Matossian tells The National.

The power of AI

The Bold Glamour effect not only simulates make-up techniques, such as applying foundation or bronzer, but also adjusts facial symmetry, from thinning the nose to chiselling the jawline – arbitrary features that are widely considered attractive.

What makes it remarkable, however, is how natural the face still appears to be, which is unusual when it comes to the social media filters popular on Instagram and Snapchat.

“Artificial intelligence and augmented reality filters are all about machine learning, computer vision and graphics overlays,” says Alfred Manasseh, co-founder of Dubai-based metaverse developer Shaffra.

“The technology first maps the face using a facial-detection algorithm that identifies key facial landmarks like the eyes, the nose and the mouth.”

Manasseh says this process results in a “digital skeleton”, which is then laid on the user's face.

The reason why the filter does not distort is because of machine learning, adds Manasseh. The filter uses “vast amounts of facial data to recognise and predict movements, which ensure the filter moves naturally”.

Salih Ismail, the discipline lead of information technology at Murdoch University in Dubai, says the TikTok filter also likely uses an advanced AI technique called generative adversarial network (GAN).

“In simple terms, GAN takes your image as input, matches it with hundreds of thousands of similar images, and generates an enhanced version of you for the best results,” he explains.

This means people who use it are not just looking at their own faces, but rather a seamless combination of other facial features mashed into one polished look.

Ismail says this application of AI is “at a tipping point”, and that advanced image manipulation could result in a future “where the line between real and enhanced is thin”.

Blurred lines

Understanding the technology behind Bold Glamour is one thing, but pondering its mental health implications is another.

Meta – which owns Instagram and Facebook – and TikTok's parent company ByteDance, have recently come under fire for their increased involvement in many aspects of people's lives, from politics to mental health.

Dubai resident Patricia Toledo admits she spends the majority of her day scrolling through TikTok. The first thing she does when she wakes up is reach for her phone, open the social media application and scroll for a few minutes before starting her day.

And she repeats this throughout the day – in between work, in the bathroom and just about every time she finds herself idle.

“Sometimes, even when I don't even find videos that are entertaining, and even when I'm not completely paying attention, I just scroll. It's addictive,” says Toledo.

Toledo is one among many millions of people around the world who actively use social media.

Facebook, for instance, has three billion monthly active users, according to recent data from Statista.

TikTok has achieved the fastest growth in users. Since its 2016 launch, the app has reached a billion monthly active users, which is half of the users of Instagram, which launched in 2010.

The rapid rise of TikTok has led the platform to roll out several features to make the online community a safe space for users.

“TikTok's diverse community transcends generations, spanning from teenagers to grandparents and everyone in between,” a spokesperson for the Middle East and North Africa market tells The National.

“We build our community guidelines with these different audiences in mind, by limiting features by age, empowering our community with content controls, and supporting families with parental controls.”

In addition to the guidelines, the platform introduced a new content classification system last year. The algorithm-powered feature organises videos uploaded on the platform based on thematic maturity “to prevent content with overtly mature themes from reaching younger audiences between the ages of 13 and 17“.

The system allocates a maturity score to videos, which are then filtered to only appear on screens of those who are eligible to view them.

In terms of the concerns around Bold Glamour and other beauty filters, TikTok has a guideline that is meant to ensure that effects, especially those created by third parties, comply with its community policies.

“When we identify an effect that violates these policies, we remove it,” says the company, adding that users can also report filters that they think are in violation of their guidelines.

Matossian believes these regulations are crucial, especially because platforms such as TikTok can only have so much control when it comes to how people use them. She recognises how online communities can also offer some benefits to users.

“There are several communities on social media platforms that promote an engaging and constructive outlook when it comes to body positivity, mental health awareness and autism acceptance, to name a few,” she says.

She adds that the use of filters is not always negative, but what's more important is to understand why people use them in the first place.

“Thoughts, behaviour and feelings are underpinned by a deep human need for belonging. Underneath the layers can be a lack of satisfaction, self-worth, or self-identity in searching for a better version of ourselves,” Matossian explains.

“This awareness may lead to pursuing ways of feeling better and looking better.”

Toledo, who says she uses beauty filters once in a while in her social media posts, says the enhancements sometimes offer a quick self-esteem boost. However, she is aware that relying on filters for confidence is problematic.

“After all, we still spend a lot of time in the real world, and these filters won't really help.”

How to build mental resilience when online

As social media continues to play a major part in people's lives, Matossian offers some tips to build mental and emotional resilience.

• Focus on what your body can do, not what it looks like
• Cultivate a sense of self-worth where validations come from within, not from others
• Have a realistic awareness of what you see online; be honest and admit that it is not representative to compare yourself to what is not real. Differentiate fact from fiction
• Give yourself space to contemplate and reflect when online. Be curious about content
• Validate your body by asking it what is sees, feels and hears. Try to listen to your body, not change it.