A recent study provides a detailed analysis of TikTok’s popular “Bold Glamour” filter, finding that its artificial intelligence makes specific, subtle changes that often align with the goals of common cosmetic procedures. The research, published in Plastic & Reconstructive Surgery-Global Open, quantifies these modifications to better understand how such digital tools are shaping modern beauty standards and influencing patient expectations in aesthetic medicine.

The Bold Glamour filter, released on the TikTok social media platform in early 2023, quickly gained widespread attention for its remarkable realism. Unlike earlier filters that often glitch or appear as a simple layer over a person’s image, Bold Glamour uses a more advanced form of artificial intelligence to generate a new, altered version of the face in real time.

While TikTok has not revealed the specific technology, experts suggest it employs a process known as a generative adversarial network, which compares a user’s face to a vast database of other images to produce a completely remodeled, airbrushed result.

This technological sophistication means the digital enhancements remain stable even when a user moves or covers parts of their face, making the effect appear uncannily natural and often difficult to detect. Researchers undertook this study because the filter’s seamless and persuasive nature raises questions about its impact on self-perception.

Psychologists have expressed concern that such realistic filters could warp a person’s understanding of what a normal face looks like, potentially making them feel alienated from their own appearance and more interested in pursuing plastic surgery. The filter’s ability to present a glossier, more sculpted version of oneself has prompted a broad public conversation about its potential to establish new, algorithmically-defined benchmarks for beauty that could influence requests for aesthetic surgery.

To measure the filter’s effects, the research team captured images of 10 female participants using a standard smartphone camera. For each participant, they analyzed one unfiltered photograph and one image generated by the Bold Glamour filter. The analysis was conducted using two methods. First, they employed an automated software that can identify and map nearly one thousand distinct points on a human face, allowing for precise digital measurement of changes to facial features. Second, the researchers performed manual observations to identify alterations in soft tissues, such as skin texture and the appearance of makeup.

The analysis revealed a pattern of distinct and consistent modifications. On a broad scale, the filter tended to increase the face’s overall dimensions, making the jawline appear wider and the forehead taller. Many of these structural changes were proportionally small, often altering a feature by less than 10 percent. This subtlety helps the filtered image retain a natural appearance while still creating a noticeable enhancement.

When examining specific features, the researchers documented more pronounced changes. The filter algorithmically refined the nose, reducing the width of the tip and the base while increasing the upward rotation of the nasal tip. It also created substantially fuller lips, augmenting the height of the upper lip by an average of 25 percent and the lower lip by 16 percent. Other changes included a slight lift of the right eyebrow and a more defined projection of the cheekbones.

Perhaps the most significant alterations were not structural but related to the skin and soft tissues. The filter produced a dramatic smoothing effect, which reduced the appearance of wrinkles, tear troughs under the eyes, and smile lines around the mouth. It also seemed to eliminate hyperpigmentation and apply digital makeup, such as thickening and darkening the eyebrows.

These changes mimic the results of common nonsurgical rejuvenation treatments, including chemical peels, laser resurfacing, and microblading. The filter appears to blend both feminizing features, like lifted eyebrows, and masculinizing ones, such as a stronger jaw, likely to appeal to a wide range of users.

The researchers note that these findings have direct relevance to the field of aesthetic medicine. Digital tools like the Bold Glamour filter could one day serve as aids during consultations, helping surgeons and patients visualize potential outcomes of procedures. For instance, the filter’s effects correspond to the goals of treatments like lip and brow lifts, rhinoplasty, and various skin resurfacing techniques. Such technology could facilitate better communication and help establish clearer expectations between a patient and a surgeon.

At the same time, the widespread use of these filters presents challenges. The digitally perfected images may promote unrealistic beauty ideals that are difficult or impossible to achieve through surgery. This could affect the psychological well-being of users, potentially contributing to body image issues or conditions like body dysmorphic disorder, where a person becomes obsessed with perceived flaws in their appearance. The study suggests that surgeons must be mindful of these risks, balancing the potential benefits of new technologies with an ethical responsibility to prioritize the patient’s overall health and well-being.

The study was not without its limitations. The analysis was based on a small sample of 10 female participants, so the findings may not apply to men, nonbinary individuals, or a more diverse population. Additionally, the photographs were taken with a smartphone rather than under standardized studio conditions, which may have introduced minor inconsistencies in the automated measurements.

Future research could address these points by using larger and more diverse participant groups and employing professional photography to ensure greater precision. Further investigations are also needed to understand the psychological effects of prolonged interaction with AI-modified self-images, especially among younger and more vulnerable populations.

The study, “Artificial Intelligence Beauty Filters and Aesthetic Surgery: Insights from TikTok’s Bold Glamour Filter,” was authored by John A. Toms, Annie M. Fritsch, Elizabeth O’Neill, Jubril Adepoju, and Mamtha S. Raj.

A new study has found that a person’s political affiliation is a powerful factor in online dating choices, carrying about as much weight as physical attractiveness. At the same time, the research suggests that a willingness to date someone from an opposing party, a signal of political tolerance, is an even more desirable trait. The findings, published in Political Science Research and Methods, provide a nuanced look at how political divisions are shaping our most personal relationships.

The research was conducted by a team from Queen Mary University of London and the London School of Economics and Political Science. They were motivated by the observation that political polarization has begun to influence decisions far outside the voting booth, from hiring to personal friendships.

The researchers questioned whether this bias is purely about party labels, or if those labels act as a shorthand for other assumed characteristics, such as values or lifestyle. By focusing on the complex world of online dating, they sought to disentangle the raw effect of partisanship from the many other factors that guide the search for a partner.

To investigate these questions, the scientists designed a realistic online dating simulation for 3,000 participants in the United Kingdom. Each participant was shown a series of paired dating profiles and asked to choose which person they would prefer to date. The profiles were generated with a mix of randomly assigned traits, creating a wide variety of potential partners. This method, known as a conjoint experiment, allows researchers to precisely measure the independent influence of each characteristic on a person’s choice.

The profiles included key political attributes, such as party affiliation (Labour or Conservative) and political tolerance. The tolerance attribute was presented as a statement in the profile’s bio, either expressing openness (“Open to match with anyone”) or intolerance (“No Tories/Labour!”). Profiles also featured nonpolitical traits common on dating apps, including physical appearance, race, education level, height, and even dietary habits, such as being vegetarian. The use of actual photographs, pre-rated for attractiveness, was intended to make the experience more similar to using a real dating app.

The results showed that political identity has a substantial effect on dating decisions. On average, a person was 18.2 percentage points more likely to be chosen if they shared the same party affiliation as the participant. This effect was similar in magnitude to the preference for a physically attractive person and was twice as strong as the preference for a potential date with a university degree. This suggests that in the modern dating market, political alignment can be just as important as conventional standards of attraction.

However, the single most influential trait was not party affiliation, but political tolerance. A profile that signaled an openness to dating people from any political background was nearly 20 percentage points more likely to be chosen than a profile expressing intolerance. This preference for open-mindedness was slightly stronger than the preference for a shared party. Participants appeared to value tolerance even when evaluating someone from their own party, indicating a genuine appreciation for the trait rather than just an aversion to being rejected themselves.

The study also uncovered a notable asymmetry in partisan behavior. While supporters of both major parties preferred to date within their own political group, this tendency was much stronger on the left. Labour supporters were approximately twice as likely to choose a fellow Labour supporter compared to the rate at which Conservatives chose other Conservatives. This finding points to different social dynamics within the two partisan groups in the UK.

Another surprising asymmetry emerged when participants encountered profiles that defied political stereotypes. Conservative participants were more likely to select a Labour supporter who broke from the typical mold, for example, by being White or holding “traditional” values.

In contrast, Labour supporters were less likely to choose a Conservative profile that broke stereotypes, such as a Black or vegetarian Conservative. The researchers suggest this could be related to a negative reaction against individuals who violate strong group expectations, making them seem unfamiliar.

The researchers acknowledge certain limitations. The study focused only on Labour and Conservative supporters, which may not capture the full complexity of the UK’s multiparty political system. While the experiment identifies these differing preferences between partisan groups and genders, it does not fully explain the underlying psychological reasons for them. Future research could explore these motivations in greater depth.

Additional work might also examine the role of geography, as dating pool size and composition in urban versus rural areas could alter how people weigh political and nonpolitical traits. The influence of other major political identities, such as a person’s stance on Brexit, could also be a productive area for investigation.

The study’s findings suggest that while partisan divides are real and affect relationship formation, they are not absolute. An expressed sense of tolerance may be one of the most effective ways to bridge these political gaps in the personal sphere.

The study, “‘Sleeping with the enemy’: partisanship and tolerance in online dating,” was authored by Yara Sleiman, Georgios Melios and Paul Dolan.