Deepfakes: When reality becomes a digital illusion

Deepfakes have the potential to influence elections by spreading misinformation and manipulating public opinion. Although concerns about this were expressed a few years ago, we are starting to see tangible consequences. As recent examples, according to Radio New Zealand (RNZ), New Zealand’s public broadcaster, the country’s national party has used artificial intelligence to produce political material ahead of next October’s general election.

Also in the United States, deepfakes are widely used in the political context. Already in 2018, a group of digital artists made a deep fake video showing former US President Barack Obama, giving the illusion that he was making shocking statements about Donald Trump. Today, with the next presidential election already under way, the GOP is producing AI-generated videos attacking Joe Biden.

More recently, a manipulated video clip suggested that Ukrainian President Volodymyr Zelensky was ordering his soldiers to lay down their arms and give up the fight against Russia. Ukraine has frequently warned its people about the possibility of Russia broadcasting manipulated videos, to create panic and confusion.

social repercussions

Deepfakes can be used for entertainment. It can also pose a great danger to individuals and society as a whole. In addition to their potential political use, they aid scams and scams, by means of identity theft or creating false identities, to carry out financial fraud, identity theft or even extortion. In 2020, a Geneva-based Swiss company specializing in raw materials was defrauded using deepfakes. Scammers have created fake videos of senior company officials validating major business transactions. The deepfakes were so realistic that they deceived the business partners and banks involved, resulting in fraudulent money transfers of great value and resulting in huge financial losses to society. A similar case was reported in 2019 where a UK-based energy company was targeted by fraudsters who used a voice message to mimic the voice of the company’s CEO, resulting in a fraudulent transfer of around $243,000.

Another potential consequence of deepfakes is breach of privacy. It is possible to create a deep fake video of a naked person from stolen or non-existent photos or videos, opening the door to blackmailing or defaming the person. In 2021, there were reported cases of deepfakes in California, where individuals’ faces were superimposed over pornographic content without their consent. These deepfakes were shared on online platforms, resulting in severe consequences for the victims, both personally and professionally.

The importance of awareness

Frequent deepfakes cases show the importance of public information and prevention measures.

Major technology companies such as Google, Microsoft, and Meta have publicly condemned such use of deepfake technology and are developing tools to detect it. Microsoft is creating a new anti-deepfake technology, called Microsoft Video Authenticator.

In Switzerland, researchers at the École polytechnique fédérale de Lausanne (EPFL) in collaboration with cybersecurity firm Quantum Integrity (QI), have developed software that uses statistical learning (machine learning) to detect manipulated images, including deepfakes. The software learns to distinguish between true and false using the same process used to create deepfakes.

It is also necessary for legislative authorities to take action. In April 2021, the European Commission proposed a legal framework to regulate the use of AI which aims to establish clear rules for high-risk AI systems, including deepfakes, to ensure their transparency, accountability and safety.

The combined efforts of technology companies, legislatures, and informed citizens are necessary to thwart the dangers of deepfakes.