Since 1975
  • facebook
  • twitter

Seeing isn’t believing: AI Summit’s warning on deepfakes

COVID-19 has fueled misinformation, with individuals using social media algorithms to spread false information. (AFP/File)
COVID-19 has fueled misinformation, with individuals using social media algorithms to spread false information. (AFP/File)
Short Url:
13 Sep 2024 03:09:24 GMT9
13 Sep 2024 03:09:24 GMT9

Sulafa AlKhunaizi

RIYADH: The use of deepfakes — images or recordings that have been altered or manipulated to depict real or nonexistent people — raised worrying ethical questions, but could also bring positive, life-changing benefits, the Global AI Summit in Riyadh was told on Thursday. 

In a panel titled “Deepfakes: Navigating Ethical, Technical, and Social Implications” Thomas Schneider, ambassador and director of international affairs at the Swiss Federal Office of Communications, said that definitions of a “deepfake” can have a political, technical, or legal basis.

The technology has deep ethical implications depending on “whether it’s a piece of art or (if it is) insulting someone.” 

On the same panel, Jennifer Williams, assistant professor at the University of Southampton, said that deepfakes could have “minuscule edits, minor alterations.” 

“It only takes three to five seconds of your voice in order to create a deepfake that can bypass automatic speaker verification systems that we use, for example in the UK,” she said.

Williams said that deepfakes are not always designed to be misleading or illicit.  

The rise of speech synthesis technology, a form of deepfake, is leading to growing demand in a number of companies, she said.  

Benefits of the technology include use in voice reconstruction, voice privacy that involves sourcing biometric information from the speech signal, and screening readers for those who are visually impaired through speech synthesis technology. 

Williams warned that deepfake technology could also lead to a rise in mental health issues due to fake media consumption.  

“One of my concerns is that we might see various new types of mental health issues surrounding how we question our reality. I don’t know if the younger generation is going to come up to speed as quickly as we have with the changing world and how that might impact how they handle information or just each other,” she said.

Walid Magdy, co-founder and CTO at Alan AI, said that deepfakes have been put to negative use on social media platforms, such as a TikTok account operated by a man impersonating US actor Tom Cruise.  

The slow response in removing deepfakes from platforms poses problems, he said. 

“To convince the people to remove the harm will take even longer. It’s the same with fake news — it propagates among people, and they think it’s true and act on it. 

“Don’t believe everything you see; you have to double check many times.” 

At the end of the panel, Schneider discussed the future of AI, saying: “We need to learn and step up. It’s a mix of measures. We need to technically develop standards that help us. We need to legally develop standards that help us. We need to develop our own skills and common sense. We have no choice.” 

topics
Most Popular
Recommended

return to top

<