Deepfakes have become a tool used by cybercriminals, hacktivists, fake news outlets and others faster than expected, according to a new report
Released on Thursday by the deepfake-monitoring company Sensity, the report looks at the state of different threats posed by the technology.
“Cybercriminals, hacktivists, adversarial countries, fraudsters, fake news outlets, and cyber soldiers have quickly incorporated AI technologies into their attack and deception frameworks, faster than anyone in the public and private sectors expected,” the intro to the report read.
Founded in 2018, the European company Sensity specialises in deepfake detection.
They used anonymised data from their clients to gather insights on the risks posed by deepfakes in 2023 and the first half of 2024.
The first point that the report highlighted is the increasing sophistication of the technologies and the plethora of tools to make them that are available.
Politicians are the most mimicked category
Politicians represent nearly 40 per cent of the people targetted by deepfakes, followed by celebrities who represent nearly 30 per cent.
Businesses, meanwhile, represent nearly 20 per cent of the hyper-realistic imitations.
Politicians are mostly targeted by imitations of themselves making false statements to either influence elections or swing public opinion.
For example, the report gave the example of a deepfake of a Ukrainian politician saying in a fake TV interview that the country was behind a terrorist attack in Moscow.
“Although the election campaign is still in an early phase we have found initial evidence of deepfake weaponisation during the primary election in particular against the main Donald Trump opponents,” noted the report regarding the upcoming US elections.
Scams are increasingly involving deepfakes
Celebrities and businesses are often victims of deepfakes used in scams.
A recent deepfake led to a transfer of €23 million to fraudsters, for instance.
Several celebrities such as Tom Hanks, Elon Musk or YouTuber MrBeast have been impersonated to promote scams.
Usually, scams are shared on social media because of the viral potential and the possibility of having targeted advertising, enabling scammers to reach specific groups more likely to be interested in trading, gambling, or crypto, according to the report.
Trading is by far the most targetted industry by deepfake scams, representing around 35 per cent of them, followed by retail and gambling, both at around 15 per cent. Public subsidies scams represent 12.5 per cent.
The report also underlined high-tech scams that involve bypassing biometric verification to access online banking or financial services.
The market of catching deepfakes
Many are playing catch up with Big Tech now developing new techniques to spot AI-generated photo and video content.
Sensity has its own tool for analysing the pixels and the file structure to detect if it has been modified.
Intel launched a real-time deepfake detector in 2022 that inspects how light interacts with blood vessels in the face.
Meta’s platforms will also soon label AI-generated content for their users.
