- June 25, 2022
- By Cyberarch Admin
For a long time, since the advent of advanced technology-use for business purposes around the world, audio and video records were considered reliable and safe. But that’s not the case anymore. With the rapid increase in the generation of synthetic video and audio and the use of deepfake technology, the authenticity of digital reality is at stake. Fake media has the potential to wreak havoc online.
A decade ago, trusting a phone call from a business associate or a dear friend was easy. Similarly, the authenticity of videos featuring well-known personalities such as politicians or celebrities was never in question. However, the sophisticated technology and evolving AI have made deepfake technology a serious concern today, subverting everything people once believed. Deepfake videos and audios are developed with scary realism, and it threatens to reach and disrupt personal and political spheres.
In addition, deepfakes have started to emerge as a major challenge for cybersecurity measures on a global scale, threatening to cause huge economic loss. Businesses and governments must be wary of the impact and emphasize taking countermeasures right now.
In this article, we explore many emerging facets of deepfake technology, and how it is affecting or can affect individuals, businesses and governments in the future. We also take a look at possible solutions as far as cybersecurity is concerned. Read on to know more.
Deepfake, as the name indicates, uses deep learning (machine learning) and artificial intelligence to create synthetic media, which includes fake videos and audio. In other words, machine learning algorithms are used to create people, voices and images that do not actually exist.
The deep learning model helps in imitating the voice or image of a person which is exploited by threat actors to conduct various malicious activities, mislead people, spread false information in the public domain and other nefarious purposes.
Deepfakes can be used for different purposes such as:
- Celebrity Pornography
- Influencing Elections
- Social Engineering
- Identity Theft
- Financial Fraud
- Disseminating Misinformation
Deepfakes: Past and Present
Deepfake technology and the associated deepfakes phenomenon might seem relatively new to you, however, it isn’t. The technology is built on inventions of the last century such as the computer, software, algorithms and lately artificial intelligence.
Although it is difficult to pinpoint a person or the moment when deepfake was invented, a video rewriting program that generated fake audio was the possible beginning. Further research on such technologies using Artificial Intelligence was done but remained unpopular.
Finally, in 2017, the term deepfake was coined by a Reddit user and it became popular. The user with the same name used open-source face-swap technology to share adult celebrity deepfake videos.
Generative Adversarial Networks (GANs) is often regarded as the key component of deepfakes. Developed in 2014, deepfakes as people know them today, are the direct result of the possible misuse of the machine learning technology.
This technology consists of two AI programs. One AI program generates content like images, audio or video. The second one performs the function of guessing if the content is real or fake. Initially, the generating AI begins with real ideas and detecting AI can distinguish original from duplicates. However, they become better as they progress, leading to the generation of fake media with a high amount of realism.
The State of Deepfakes: Statistics
Since 2017, the deepfakes have exponentially increased all across the world. According to reports, there are more than 14,000 deepfake videos available online with over 100 crore views. As per experts, the majority of these are adult-oriented videos online, covering almost 96% of deepfakes in the digital world. However, lately, non-pornographic deepfakes are increasing as well, which is a cause of worry.
Deepfake Attack Types
There are different types of deepfake attacks taking place worldwide.
- Real-time – In real-time deepfake attacks, a fake video or audio is created to fool or defraud the target. The victim, which can be a client or a colleague, ends up believing in the deepfake media. Generally, in this type of attack, the criminal behind this attack shows the urgency to create panic and make the victims react fast without considering consequences.
- Non-real time – In this type of attack, video or audio is shared through various asynchronous communication channels. This reduces the chance of suspicion among the target.
- Fraud– Other attack type includes ghost fraud where cyber criminal steals the data of a deceased person and utilize it to make financial gains. Also, fake bank accounts are created using stolen identities which leads to massive financial damage. Furthermore, synthetic identity fraud is carried out where information and credentials of multiple people are stolen to create a new persona and make financial gains.
- Phishing – Phishing attacks are carried out through mail, online communication platforms and other mediums. A well-timed deepfake video or audio is used to compromise businesses, leading to financial losses.
- Biometrics – Identity theft is one of the major challenges faced by cybersecurity leaders working for various businesses across verticals. Biometrics is one of the efficient methods to boost security and enable Zero Trust. However, deepfakes threaten to compromise security based on biometrics and make organizations vulnerable to cyberattacks.
Deepfake as a Service
In 2018, Reddit removed the Deepfake user. However, the damage has been done. The growth of AI in technology has already led to progress in deepfakes. Consequently, the potential to use deepfake as a service and related capabilities among threat actors have multiplied enormously in the last many years.
This also enables cybercriminals to access such advanced technologies easily and at a cheaper price in the future. Social engineering attacks may become more commonplace and more profits can be made from victims using deepfake technology.
Additionally, the research on GANs is increasing at a fast rate as well. New capabilities are being discussed and developed to improve and support real-world applications. There are communities and forums online disseminating valuable information to the masses throughout the world. People interested in creating deepfakes can easily connect with experienced developers now.
Open-source code repository such as Github has face swap source code which can be accessed by anyone. Moreover, programmers regularly contribute to making it better, efficient and usable. Various computer apps have been made available with face swap capabilities and voice cloning. Service portals are open online that provide customized deepfake media. The growing interest in deepfakes is terrifying, to say the least.
Without a doubt, this threat vector is gaining momentum and may emerge as one of the biggest challenges for cybersecurity professionals and businesses in the near future.
Deepfakes and Businesses
In 2019, the first known deepfake attack was conducted. The CEO of a UK firm got a call, which the victim believed was his boss, asking him to transfer 243,000 dollars to a Hungarian supplier. The CEO found the head of the firm’s parent company to be genuine with the right tone and German accent. However, it was realized that he was tricked by deepfake technology.
In 2020, a tech firm was targeted but unsuccessfully. In a 2021 report from Recorded Future, it was observed that threat actors are increasingly tilting towards deepfakes to conduct cyber crimes.
The new work-from-home culture and distributed workforce have opened up opportunities for cybercriminals to attack businesses globally.
Deepfakes and Politics
The use of deepfakes can have major ramifications in the political discourse of a country. In the past years, many cases have come to light where deepfake videos and audio were allegedly used.
On the other side, more awareness about deepfake videos has given rise to doubts and suspicions even about original and true videos. In the future too, deepfakes can greatly influence politics, and elections in nations, stifle democracy and may cause irrevocable damage.
Some of the examples are discussed below.
In 2018, the Gabonese President Mr. Ali Bongo disappeared from public life for a brief time, leading to massive speculations and unrest among the political class and common people. The government thus released a video of the President addressing the nation. However, the abnormal appearance of the President led many in the online world to question its authenticity. It ultimately led to a military coup. But the forensic analysis did not find any evidence of deepfakes and the President made public appearances later too.
In 2019, a political scandal came to light in the form of a tape where the Malaysian Minister of Economic Affairs was caught having sexual activity with an opposition minister’s aide. Since then, it has been argued that the video was a deepfake. However, forensic analysis has found no signs of manipulation.
Shallow fakes is another phenomenon fast emerging in the online world. It refers to videos which are edited and manipulated to take the speech of the person out of context. This is done to tarnish the reputation of the person. In 2019, a shallow fake video of Nancy Pelosi of the US was made which went viral online.
Election Security Challenges
Deepfake is fast becoming a major challenge that threatens to upend established security for free and fair elections. More and more experts are convinced that deepfakes will be used in disinformation campaigns during elections and influence voters.
Fraudulent information regarding candidates, polling sites, voting dates and processes can impact election outcome. Moreover, fake and misleading data can lead to chaos and confusion in elections. Democratic elections value integrity more than anything else and deepfakes are a clear challenge to it today.
Deepfakes and Cybersecurity Challenges
Deepfake threats are already changing the security landscape by enhancing and enabling sophisticated cyber crimes such as extortion, fraud and manipulation. Fake digital identities are forged to conduct fraud, infiltration and even espionage.
According to reports, multiple cases of fake profiles on social media platforms were found impersonating people and manipulating and extracting information. Experts after analysis found that the images were developed synthetically.
Also, as mentioned before, biometric security based on iris scan, fingerprints, voice and face recognition face a severe challenge from deepfakes. Thus, it is high time businesses, governments and individuals find appropriate solutions to mitigate risks and boost protection again deepfakes.
Security leaders can play a key role in implementing effective solutions across the business.
Training and Awareness
Ongoing training should be provided to the team regarding potential deepfake-related threats, the key components and how they can affect businesses, and what are the best possible solutions.
Awareness sessions can teach employees how to detect deepfakes such as unblinking eyes, abnormal facial movements and odd lighting or colour. This can help everyone to be alert and prepared.
In addition, the team members can be trained to be wise when receiving any urgent wire transfer requests even from the heads of business. This can help employees to be alert and further authenticate such requests.
Another effective way to challenge deepfakes is by enabling employees to challenge any suspicious activity or demand from higher authorities. Verbal verification questions can also be asked to make sure the communication is reliable.
Software-based tools can be used by the IT team to verify and detect if the video or audio is reliable or not. Artificial Intelligence tools can also help in identifying fake videos and other media.
Action Plan & Policy
In case of a deepfake attack, an action plan with the right response must be in place to minimize the loss and effectively deal with the problem as soon as possible. Moreover, a robust security policy and its implementation can help everyone in the organization to take necessary steps whenever any security concerns arise.
Partner with Cybersecurity Companies
Also, you can always partner with cybersecurity companies who will take necessary measures and use the latest tools to counter deepfakes from harming your business.