The rise of sophisticated false profiles on social media platforms has become an increasingly urgent issue, especially with the role of AI-produced synthetic and text-to-image generated media. In response to the growing problem, LinkedIn has partnered with UC Berkeley to develop cutting-edge detection methods to combat numerous fake profiles.
A ground-breaking detection method, boasting a 99.6% accuracy rate in identifying artificial profile pictures and a mere 1% misidentification of genuine images, brings forth a new era in AI-generated media detection.
Forensic methods used to investigate AI-generated profiles come in two primary types: hypothesis-based methods and data-driven methods (machine learning). Hypothesis-based methods excel in spotting oddities in synthetically made faces and learning semantic outliers. However, with the rapid advancements of synthesis engines, this approach faces limitations.
Data-driven methods, on the other hand, differentiate natural faces from those created by computer-generated imagery (CGI). Although powerful, these machine learning-based methods struggle with classification when encountering images outside their area of expertise.
To address these limitations, a proposed hybrid approach consists of two crucial steps: identifying a unique geometric attribute in computer-generated faces and employing data-driven methods to measure and detect it. This innovative technique allows for a lightweight and quickly trainable classifier, requiring training on only a small set of synthetic faces.
In order to test and validate the new approach, 41,500 synthetic faces were built using five distinct synthesis engines and compared to 100,000 real LinkedIn profile pictures. This comprehensive comparison provided valuable insight into the similarities and differences between real and synthetic faces.
Notable observations include the diversity in real photos, contrasted by the clear features and standardized ocular location and interocular distance found in StyleGAN faces. Another intriguing difference revealed was in typical photo composition, with real photos often featuring upper body and shoulders, and fake photos focusing on the neck-up area.
To further refine the detection technique, a one-class variational autoencoder (VAE) combined with a baseline one-class autoencoder was employed, focusing on synthetic faces rather than face-swap deepfakes. The result was a simpler, easier-to-train classifier with comparable overall classification performance.
An evaluation of the generalization ability of this technique was conducted using images generated by Generated.photos and Stable Diffusion. Findings indicate that the method was relatively generalizable for GAN-generated faces from Generated.photos, but not generalizable for faces from Stable Diffusion.
In conclusion, the proposed hybrid technique offers a promising avenue to identify AI-generated synthetic profiles on platforms like LinkedIn. The benefits and potential applications of this research could help safeguard online spaces by improving our ability to detect fake profiles. As we continue to rely increasingly on our digital presence for personal, professional, and business purposes, staying vigilant for fraudulent profiles remains a critical task for ensuring the integrity and overall safety of our online networks.
Up until working with Casey, we had only had poor to mediocre experiences outsourcing work to agencies. Casey & the team at CJ&CO are the exception to the rule.
Communication was beyond great, his understanding of our vision was phenomenal, and instead of needing babysitting like the other agencies we worked with, he was not only completely dependable but also gave us sound suggestions on how to get better results, at the risk of us not needing him for the initial job we requested (absolute gem).
This has truly been the first time we worked with someone outside of our business that quickly grasped our vision, and that I could completely forget about and would still deliver above expectations.
I honestly can't wait to work in many more projects together!
![The ‘Giveaway Piggy Back Scam’ In Full Swing [2022]](https://www.cjco.com.au/wp-content/uploads/pexels-nataliya-vaitkevich-7172791-1-scaled-2-683x1024.jpg)
Disclaimer
*The information this blog provides is for general informational purposes only and is not intended as financial or professional advice. The information may not reflect current developments and may be changed or updated without notice. Any opinions expressed on this blog are the author’s own and do not necessarily reflect the views of the author’s employer or any other organization. You should not act or rely on any information contained in this blog without first seeking the advice of a professional. No representation or warranty, express or implied, is made as to the accuracy or completeness of the information contained in this blog. The author and affiliated parties assume no liability for any errors or omissions.
