An Emirati researcher is developing a platform to detect “deep fakes” in photos, videos, and audio clips

Emirati researcher specializing in cybersecurity, Bashayer Al-Salami, has developed a platform based on artificial intelligence techniques to monitor digital manipulation of images, video clips, and audio materials, in light of the growing risks of “deep fake” technologies and their widespread use in producing content whose authenticity is difficult to verify across digital platforms and social media sites.

Researcher Bashayer Al-Salami designed a platform called “Debbie,” based on a research project she prepared during her master’s studies at Zayed University, which focused on modern digital threats and the rapid development of tools for manipulating content using artificial intelligence, at a time when modern technologies have become capable of producing images, videos, and audio clips that appear realistic to a degree that may exceed the ability of the average user to discover them.

Al-Salami told Emirates Today that the development of artificial intelligence has created new digital challenges, the most prominent of which is “deep fakery,” which allows modifications to images, video clips, and audio materials in ways that the recipient may not be able to notice easily, which increases the difficulty of distinguishing between real content and digitally modified content.

She explained that the “Debi” platform works by analyzing the files uploaded by the user, then issuing an assessment of the level of reliability of the content, noting the possibility of using artificial intelligence tools to manipulate the digital material.

She added that the platform relies on advanced algorithms to analyze images, videos, and audio clips, by examining technical indicators within the digital file itself, including pixel patterns, lighting differences, and distortions related to movement or sound, which are details that may not be clear during normal viewing or listening.

She pointed out that the detection process does not depend on a single indicator, but rather on a set of precise technical signs, including asymmetry in lip movement or facial expressions, unnatural variations in shadows and lighting, in addition to minor distortions at the level of frames or pixels, as well as “digital fingerprints” left by algorithms for generating images and audio content.

Al-Salami explained that the platform relies on “deep learning” models that have been trained on thousands of samples of real and fabricated content, allowing it to recognize and discover patterns of digital manipulation at early stages.

She stated that the “Debi” platform targets a number of entities and sectors whose work depends on verifying the authenticity of digital content, including government and security agencies, media institutions and journalists, in addition to companies and entities concerned with digital security and social media platforms. It also revealed work on developing customized versions for individual users, allowing anyone to verify photos or videos before resharing or circulating them, in a step aimed at expanding the scope of benefiting from digital detection tools outside specialized institutions.

Al-Salami stressed that periods of crises and political tensions usually witness a significant increase in the circulation of fabricated content, as “deep fake” techniques are sometimes used to raise anxiety or influence public opinion, which increases the need for rapid verification tools that help journalists, official bodies, and users examine digital materials before republishing or circulating them.

She added that the role of technology should not stop at the borders of detecting digital tampering, but rather extend to enhancing digital awareness and supporting the building of a more reliable and secure media environment in the digital age.