AI’S Double-edged Sword: The dangers and implications of deep fakes

Imagine a world of uncertainty where no one believes no one and distinguishing between the real and the fake is impossible. A doctored video clip has enough power to sway public opinion, manipulate young brains, target politicians and famous personalities, endanger democracy, and thus cause societal chaos on a large scale. The challenges that technological advancements have brought about become a curse for humanity and a blessing for brutality. The unfettered seepage from the judicial system permits anyone to misuse anyone’s digital property. This imagined world of chaos is not much different from our reality where the existing problem due to deep fakes is the tip of the iceberg.

With the advancements in artificial intelligence, the chances of harm caused by fabricated videos are higher than ever. You must have come across such fabricated content that seemed real but hard to believe. If yes, you might have been exposed to a deep fake. A deep fake is a video, audio, or picture, manipulated using deep learning, a branch of AI. With the availability of more computing power over the years, machine learning algorithms have become more sophisticated, increasing the quality of deep fakes. Throughout history, generative adversarial networks fueled the evolution of deep fakes.

Like others, deep fake technology can also be used for both good and evil purposes. Regarding the betterment of mankind, deep fake technology has use cases in industries such as healthcare and entertainment. “During the coronavirus pandemic, it was difficult to diagnose the diseases that arise from coronavirus infection. It was due to the lack of X-rays, CT scans, and MRI images and the resources to diagnose whether the patient had the disease or not. Here came the use of deep fake technology; computer scientists first produced deep fake images with the help of artificial intelligence and gave them to artificially intelligent models for training. The models were able to compare the deep fake images and that of the patient to diagnose whether he had the disease or not. Moreover, training artificially intelligent models on people’s data can create privacy concerns and accuracy problems. To tackle these challenges, realistic synthetic data is produced using deep fake technology.

Deep fakes can have a severe impact on the public. Through such content, bad actors can spread misinformation to fulfill their evil ambitions. It may include having illegal financial gains, generating more clicks, or igniting social unrest by misleading the masses. Politically doctored deep fakes can also pose an ominous danger to today's democratic landscape. Malicious actors can use deepfakes to sway public opinion about a specific politician. Moreover, recent text-to-video tool ‘Sora’ has also added fuel to the fire.

Identifying deepfakes is a daunting task for those without technological understanding. Nonetheless, various methods exist for identifying these deceptive digital creations. Foremost, developing a zero-trust mindset is important. Never trust anything without verifying it. Fake videos can also have many signs including differences in the skin textures and body parts, less synchronization between lip movement and voice, abnormal blinking patterns and unusual facial expressions, etc.

Governments and big tech giants can play their roles in preventing the proliferation of deceptive content. Governments can implement legislation against the dissemination of malicious content on social media. Through social awareness and public education, the harms of deepfakes can be greatly reduced. Meanwhile, big tech giants can promote the development of robust machine learning algorithms for the detection and elimination of such content on their platforms. Additionally, investing in research and development of advanced detection algorithms and forensic tools can augment the ability to identify and mitigate the impact of deepfakes.