The Battle Against Fake News Enters The Age Of Deepfakes

The issue of fake news and the importance of thorough fact-checking has been a longstanding concern in the media, even before AI technologies emerged. What we face today is simply a new challenge to overcome.

Reputable news platforms typically rely on the expensive labor of editors and moderators to manually check the news. However, given the short lifespan of news stories, thoroughly investigating the authenticity of information is not always feasible.

The expansion of fake content

The problem of fake content extends beyond text-based news to visual content as well. As companies and creators become increasingly active on social media, sharing more photos and videos, we have seen a rise in deepfakes and other manipulations of visual and audio representations of people.

The content verification and source-checking fields offer vast opportunities for innovative technological solutions and ambitious startups. However, recent developments in video translation and dubbing services, such as Google’s Universal Translator with the lip sync feature, have raised concerns about content safety. Nevertheless, with the emergence of modern AI-powered translation and dubbing services like vidby, there is now a chance to eliminate human intervention and conscious distortion of the message, providing a more reliable solution.

Video and audio synthesis technologies have been around since the late 1990s, with advancements in voice processing and computer graphics playing a key role in their development.

Modern generative AI technologies have made creating deepfakes more accessible to even non-tech-savvy users. Deepfakes have been used for more than just entertainment or annoyance; they have also been employed for malicious purposes. For example, several years ago, Symantec discovered several cases of large-scale fraud where deepfake audio was used to mimic the voices of company executives instructing employees to transfer money to specific accounts.

One of the main challenges with deepfakes is the lack of legislation in many countries to address their creation or removal. As deepfakes become more common and realistic, the responsibility for the safe development of technology and future expectations remains uncertain.

Development of technologies for detecting deepfakes

The increasing sophistication of deepfake technology has been recognized as one of the most significant cybersecurity threats in recent years. However, the technology used to detect and expose these deepfakes is evolving in parallel.

Organizations like DARPA, Microsoft, and IBM have been working on technologies to differentiate fake content from genuine content. Deepfake software often falls short when it comes to capturing intricate details. Focusing on specific aspects such as the presence of shadows around the eyes, overly smooth or wrinkled skin, unrealistic facial hair, fake moles, and unnatural lip color can help identify deepfakes effectively.

Startups are also exploring innovative solutions to address this challenge. For example, some offer technology that embeds a digital signature into content, much like a watermark, to authenticate original photos and videos. This solution has been well-received by large insurance companies, suggesting that similar technology may soon be commonplace in cameras and mobile phones to ensure the authenticity of the content.

Many deepfake software programs are open-source, which allows for both the creation and detection of deepfakes.

As deepfake technology continues to advance, so will the methods for exposing and countering them. The ongoing development of these technologies highlights the need for continued vigilance and innovative solutions to address the ever-evolving challenges that deepfakes present.

Despite the strides made in deepfake detection and prevention, much work remains to be done. The lack of established legislation around the creation and dissemination of deepfakes remains a significant hurdle. As synthetic videos and audios become more realistic and commonplace, the need for such legislation grows more urgent.

It is also crucial to consider the ethical implications of these technologies. Although AI has immense potential to transform industries and improve our lives, its misuse could have severe consequences. Therefore, the safe and ethical development of AI technologies, including deepfakes, must be a priority.

Conclusion

Research institutions, tech companies, and policymakers will need to collaborate to tackle new challenges. While technology can provide some of the tools necessary to detect and prevent the misuse of AI, education and public awareness are also critical in mitigating the impact of deepfakes. People need to be aware of the existence of deepfakes and the potential for manipulation, and they need to be equipped with the skills to evaluate the content they encounter or at least apply critical thinking to anything they read and watch.