Software Engineer Asst. Prof. Kristin Surpuhi Benli stated that video and audio content produced with artificial intelligence is now widely used in many fields such as cinema, education, advertising, gaming, and media. She said, “Commercials, online education videos, characters in digital games, virtual news anchors, and content translated into different languages are prepared quickly and at low cost thanks to this technology. On the other hand, the same technologies can also be misused for purposes such as producing fake news, committing fraud, and creating information pollution.”
Noting that artificial intelligence, along with deepfake and similar technologies, has brought about a profound transformation in the concept of “reality” in the digital age, Asst. Prof. Kristin Surpuhi Benli said, “We now have to question whether the things we see, hear, or even read are actually real.”
Asst. Prof. Kristin Surpuhi Benli from Üsküdar University’s Faculty of Engineering and Natural Sciences, Department of Software Engineering (English), evaluated artificial intelligence–supported voice and image generation technologies.
What is “deepfake” technology?
Explaining that artificial intelligence–supported voice and image generation is generally referred to as “deepfake” technology, Asst. Prof. Benli said, “Etymologically, the word ‘deepfake’ is derived from the combination of ‘deep learning’ and ‘fake.’ This technology uses deep learning algorithms developed in fields such as facial recognition, voice analysis, and image processing to produce fake videos, sounds, or images that are extremely close to reality.”
It can be used by malicious individuals
Asst. Prof. Benli stated that with deepfake technology, a person’s face, expressions, or voice can be imitated and added, “In this way, a conversation or event that never actually happened can be presented as if it really did. While this technology can be used to produce creative content, it can also be exploited by malicious individuals to spread disinformation or commit fraud.”
Fake news can be produced
Emphasizing again that video and audio content produced with artificial intelligence is widely used in areas such as cinema, education, advertising, gaming, and media, Asst. Prof. Benli said, “Commercials, online education videos, characters in digital games, virtual news presenters, and content translated into different languages are prepared quickly and at low cost thanks to this technology. However, the same technologies can also be misused to produce fake news, commit fraud, and create information pollution.”
What makes deepfake videos so convincing?
Explaining the main reason why deepfake videos are so convincing, Asst. Prof. Benli said that it lies in the increasing realism and detail that artificial intelligence technologies, particularly deep learning models, offer in the production of images and sounds. She stated, “Details such as facial expressions, blinking, lighting, and skin texture can now be imitated much more naturally. Voice cloning technologies can successfully replicate speech rhythm and intonation. This progress in image and sound quality makes fake content increasingly convincing and, as a result, makes it harder to detect such manipulations by the human eye or ear.”
The concept of “reality” undergoes a profound transformation in the digital age
Asst. Prof. Kristin Surpuhi Benli stated that with artificial intelligence, deepfake, and similar technologies, the concept of “reality” in the digital age has undergone a profound transformation. She said, “We now need to question whether the content we see, hear, or even read is truly real. This situation brings about greater skepticism in accessing information and increases the need for new methods and ethical standards that can verify accuracy and reliability.”
What are the distinguishing features in deepfake videos?
Asst. Prof. Benli noted that the techniques and methods used to detect whether an image or sound is artificially generated are continuously advancing and continued, “These detection processes are based on image processing, sound analysis, and machine learning techniques. Facial expressions and gestures are analyzed in visuals to identify fake content. Some distinguishing features that can be noticed in deepfake videos include unnatural blinking, synchronization problems between mouth movements and speech, artificial transitions in skin tone, unrealistic light reflections, and distortions in hair and edge details. In sound analysis, details such as frequency spectrum, emphasis, intonation, and breathing are examined. In addition, deep learning models trained specifically for this purpose are also used effectively in the detection of fake content.”
Advanced deepfake technologies successfully mimic even eye movements and facial expressions
Explaining that images and sounds produced by artificial intelligence have now become so realistic that they are often indistinguishable to the human eye and ear, Asst. Prof. Benli said, “Although with careful examination some artificial elements can still be detected, such as the lack of naturalness in facial expressions, inconsistencies in eye movements, or the mechanical tone of the voice, these differences are becoming increasingly subtle. Especially advanced deepfake technologies can successfully imitate many details, from facial expressions to eye movements, from voice tone to breathing patterns. Therefore, short or low-resolution fake content is often perceived as real.”
The credibility of the account sharing the video should be evaluated
Emphasizing that individuals should first investigate the source of the content when questioning the authenticity of a video shared on social media, Asst. Prof. Benli concluded her remarks by saying,
“The credibility of the account that shared the video should be evaluated, and it should be checked whether the content was published by official or reputable platforms. Comparing the video or news with other sources covering the same event is an effective method for verifying authenticity. In addition, inconsistencies between the audio and visual elements in the video should be analyzed carefully. Furthermore, artificial intelligence–based detection tools provide significant support in identifying fake content.”