Synthetic media falsehood: Seeing isn’t believing anymore

• Author:
• Author name

  Quantumrun Foresight
• February 22, 2024

Insight summary

Synthetic media, blending artificial intelligence (AI) with video, audio, and visual elements, is so realistic it's hard to distinguish from actual media. Its development traces back decades, with deep learning (DL) and Generative Adversarial Networks (GANs) playing a key role in its advancement. As this technology evolves, it presents creative opportunities and significant privacy, ethics, and misinformation challenges.

Synthetic media falsehood context

Synthetic media represents a groundbreaking combination of AI-generated content, encompassing live video, visual elements, and audio within an advanced technological framework. This media form is characterized by its exceptional realism and immersive qualities, making it nearly indistinguishable from real-world media. The creation of synthetic media can be traced back to the 1950s, undergoing significant evolution in the late 1980s and early 1990s as computational power surged. 

Deep learning is the core technology driving synthetic media, a sophisticated branch of machine learning (ML). Particularly influential in this domain are GANs, which have revolutionized the field by learning from existing images to produce entirely new yet eerily authentic ones. GANs operate using a dual neural network system: one network generates fake images based on real ones, while the other evaluates their authenticity, pushing the boundaries of what's possible in computer vision and image processing.

As AI continues its rapid advancement, the applications and implications of synthetic media grow ever more significant. While these technological strides open doors to innovation across various sectors, including video games, autonomous vehicles, and facial recognition, they simultaneously introduce pressing concerns regarding privacy and ethics. The future of synthetic media thus represents a double-edged sword, offering vast potential for creativity and innovation while challenging us to address its ethical and privacy-related implications.

Disruptive impact

A 2022 study conducted by the nonprofit Rand Corporation discusses the four primary risks of synthetic media: manipulation of elections through fabricated videos of candidates, exacerbation of social divisions by amplifying propaganda and partisan content, erosion of trust in institutions through fake representations of authority figures, and undermining of journalism by casting doubt on the authenticity of legitimate news. These deepfakes can be particularly damaging in developing nations where lower levels of education, fragile democracies, and inter-ethnic conflicts are prevalent. Misinformation is already a significant issue in these regions, and deepfakes could intensify disputes and violence, as seen in past incidents in countries like Myanmar, India, and Ethiopia. Moreover, the limited resources allocated to content moderation outside the US, especially on platforms like WhatsApp, heighten the risk of deepfakes going undetected in these areas.

Deepfakes also pose unique threats to women, given the gender disparity in pornographic content. AI-generated media has been used to create non-consensual deepfake pornography, leading to abuse and exploitation. These technologies can also pose security risks by targeting intelligence operatives, political candidates, journalists, and leaders for embarrassment or manipulation. Historical examples, such as the Russian-backed disinformation campaign against Ukrainian parliamentarian Svitlana Zalishchuk, demonstrate the potential for such attacks.

The scientific community's understanding of the societal implications of deepfakes is still evolving, with studies offering mixed results on users' abilities to detect these videos and their impact. Some research suggests humans may be better at detecting deepfakes than machines, yet these videos are often seen as vivid, persuasive, and credible, increasing the likelihood of their spread on social media. However, the influence of deepfake videos on beliefs and behavior might be less than anticipated, indicating that concerns about their persuasiveness could be somewhat premature. 

Implications of synthetic media falsehood

Wider implications of synthetic media falsehood may include: 

• Enhanced techniques in digital content authentication, leading to more sophisticated methods for verifying media authenticity.
• Increased demand for digital literacy education in schools, equipping future generations with the skills to critically analyze media.
• Shifts in journalistic standards, requiring stricter verification processes for multimedia content to maintain credibility.
• Expansion of legal frameworks addressing digital content manipulation, offering better protection against misinformation.
• Enhanced personal privacy risks due to the potential misuse of facial recognition and personal data in creating deepfakes.
• Development of new market sectors specializing in deepfake detection and prevention, creating job opportunities and technological advancements.
• Political campaigns adopting stricter media monitoring practices to mitigate the impact of fake content on elections.
• Changes in advertising and marketing strategies, with an increased emphasis on authenticity and verifiable content to maintain consumer trust.
• A rise in psychological impacts due to the spread of realistic but false content, potentially affecting mental health and public perception.
• Alterations in international relations dynamics as deepfakes become a tool in geopolitical strategies, affecting diplomacy and global trust.

Questions to consider

• How does synthetic media affect your perception of current events?
• How could the development of deepfake technology influence the balance between freedom of expression and the need for regulation to prevent misinformation and abuse?