Reality still beats deepfake technology … for now

Reuters

• Tech falls short, say experts
• Increasing use in voiceovers
• UAE tops cyberattack list

Technology experts are confident that artificial intelligence-generated deepfakes will not dominate or replace human talent.

“It sounds tremendously fake, even when it sounds ‘real’,” Ahmad Haffar, the voice behind Dubai Metro and managing partner at Mindloop Studios, told AGBI.

Deepfakes are manipulated videos, images or audio recordings created using AI.

They are increasingly commonplace in the film and entertainment industry, allowing filmmakers to make actors look younger or create clips featuring deceased performers. 

In the voiceover industry, it can reduce the time required for voice actors on a project or create voices in multiple different languages.

• Generative AI presents both opportunity and risk
• Emirates companies turn to AI in pursuit of cybersecurity
• Halting the march of malicious bots

For example, UK-based AI media generation platform Synthesia created two commercials with American rapper Snoop Dogg. It was so successful that one of the company’s subsidiaries wanted to use a version of the same advertisement. Rather than reshooting, Synthesia altered Dogg’s lip movements in the new advertisement to match the subsidiary’s name using deepfake technology.

Another well-known example is Belgian creator Chris Umé, who created Tom Cruise deepfake videos which went viral on TikTok. They were so successful that Umé started his own business where anyone can make a synthetic avatar of themselves.

The rise of deepfake has brought many benefits for businesses, including increased efficiency and productivity, but it also raises questions about the future of human talent.

Haffar believes that the technology falls short of replicating the nuances and natural aspects of a human voice.

“It is still far from achieving a legitimate human-like experience-based voice that is required across the voiceover industry,” said Haffar. “And industry professionals have no interest in this technology.” 

Ethical considerations 

The ethical implications of deepfakes and AI-generated content have also been a topic of much debate. 

The potential for misinformation, privacy concerns, and the ability to convincingly fabricate events or statements have raised alarm. 

When used without consent, as American actor Tom Hanks experienced with a fake dental plan advertisement, deepfakes can harm actors’ careers and industry credibility.

Cyber attacks using deepfake technology have surged by 450 percent in the Middle East and Africa region over the past year, with the UAE being the most targeted at 36.1 percent, according to London-based identity verification platform Sumsub.

Darknet marketplaces – anonymous platforms for trading illicit goods and services – offer deepfake creation tools. Prices for creating or purchasing deepfakes vary, ranging from $300 to $20,000 per minute, as reported at the Kaspersky Cyber Security Weekend META 2023 event in May.

Cybercriminals can steal corporate funds by creating a deepfake video or audio of a CEO requesting a wire transfer or authorising a payment. 

“Deepfakes created now for entertainment and education can be used in the future for deception and misrepresentation of actual events, and it will create a challenge for future generations to separate the truth from misinformation,” said Gaidar Magdanurov, president of Swiss data protection solutions company Acronis.

During the ongoing Israel-Hamas conflict, AI-generated images on social media spread false narratives, impacting public opinion.  

A deepfake video falsely depicting Ukrainian president Volodymyr Zelensky conceding to Russia was also spread on social media.

“This technology’s impact extends to society, eroding trust in media and institutions, potentially fuelling social conflict, and raising ethical dilemmas concerning consent, privacy, and the responsibility of creators and platforms,” said Ezzeldin Hussein, Dubai-based regional director at cybersecurity company SentinelOne.

Taking action 

To address this growing concern, experts stress the need for a comprehensive approach involving robust legal frameworks, ethical guidelines and public awareness campaigns. 

“Several Middle Eastern countries grant law enforcement authorities the right to require the removal of content that breaches the law,” said Kellie Blyth, partner at Dubai-based law firm Addleshaw Goddard.

“Regulatory authorities like the UAE’s Council for Artificial Intelligence and the Saudi Data and AI Authority have been established to create strong governance for AI more broadly, including by publishing best practice guidance.

“To date, this guidance has been non-binding, so lawmakers remain reliant on the existing law,” she said.

The future of deepfakes is likely to involve both positive and negative implications, and “society will have to adapt to it at some point, the way we adapted to cybercrime, viruses and ransomware”, added Magdanurov.