Deepfakes: A cybersecurity threat to businesses and individuals

• Author:
• Author name

  Quantumrun Foresight
• January 16, 2022

Insight summary

Deepfake technology, a blend of artificial intelligence and machine learning, has emerged as a significant cybersecurity threat, capable of creating convincingly real but fake audio and visual content. This technology has been exploited by cybercriminals to impersonate individuals, leading to potential financial losses and breaches in security systems. In response, governments, social media platforms, and cybersecurity professionals are taking steps to combat this threat, including the development of detection software, legislation, and public education initiatives.

Deepfakes cybersecurity context

Deepfake technologies became a vast cybersecurity threat to individuals and organizations in 2021. Deepfakes are now used by attackers and malicious actors to clone the voices and images of real people. In a private sector context, this new cyber threat could result in significant financial losses. 

Deepfake technology combines the words "deep learning" and "fake" by applying artificial intelligence (AI) and machine learning (ML) to create synthetic human videos or voices, as well as manipulating authentic visual and audio content. Deepfakes use deep learning algorithms, which teach themselves how to solve problems when given large data sets. Deepfakes are used to swap the faces of real people in videos and digital content to make real-looking but fake content. 

Similarly, this same technology can be applied to audio files. For example, voice mimicking has been used in a variety of applications in modern times. However, authentic-sounding voice clones could be created with AI and ML by using minimal audio samples. These voice clones have been used to create new and authentic-sounding songs by musicians who have long since passed away. Audio deepfakes have also been used in film, media, and in political messaging. However, since the first deepfakes appeared a few years ago, this technology has been considered a threat with significant implications across security, geopolitics, and culture.

Disruptive impact

Deepfakes help cyber-attackers prevail over a critical barrier, which is the difficulty of convincingly posing as another person. Deepfakes is an AI-powered technology that some cybercriminals have increasingly been experimenting with to impersonate real people by creating real looking videos or audio recordings of real people that are capable of deceiving and defrauding unsuspecting users. 

Meanwhile, cyberattacks caused by criminals compromising business emails led to higher financial losses than any other form of cyberattack in 2019. With deepfake technology, attackers can use a combination of email phishing and realistic voice messages to penetrate security systems. For example, experts report that cyber criminals tend to use effortless ways to deceive users through sending business emails with deepfake voice messages or mp4 recordings of managers or executives to convince business users to click on certain links. Unfortunately, these users are instead made to download malware.

Due to the ease of accessibility and growing sophistication of deepfakes, governments and social media platforms are being spurred to take steps for prevent deepfake usage to manipulate consumers. For example, California passed a bill, AB 730, that aims to limit deepfake usage in political campaigns. Facebook and Twitter have taken steps to combat deepfakes by announcing their initiatives (in 2021) to remove deepfakes created for malicious intent.  

Presently, techniques and software exist to detect and identify deepfakes, often by identifying repeated patterns found in the algorithms deepfakes use. However, more sophisticated deepfakes are always being built to beat these detection applications, meaning cybersecurity professionals will have to constantly innovate ahead of this technology. 

Implications of deepfake cybersecurity

Wider implications of deepfakes cybersecurity may include:

• Combating the growing number of cyberattacks on businesses and individuals by using AI to fight AI deepfakes.
• Identifying and preventing the use of malicious audio and videos. For example, the inclusion of noise pixels to analyze the acoustic spectrum or frames of media files, or to prevent modifications in videos to detect any form of distortions.
• Reducing or preventing political campaign advertising which may be deceptive to voters or injure the reputation of candidates.
• The need for advanced digital literacy programs to educate the public on discerning real from manipulated content.
• Substantial financial losses requiring the development of sophisticated fraud detection systems, which could create new job opportunities in the cybersecurity sector.
• The growth of counter-technologies designed to detect and combat deepfakes, leading to a dynamic and continually evolving tech landscape.
• A higher demand for cybersecurity professionals, resulting in job growth in this sector and the need for specialized training and education programs.

Questions to consider

• What are the possible regulations that governments can put in place to prevent the spread of deepfakes, aside from the preventing the manipulation of political campaigns?
• Will the use of deepfakes undermine trust, even though cybersecurity professionals and agencies are building software and developing techniques to detect and identify deepfakes?