A recent study performed by Synthedia has found that consumers are, by and large, ambivalent over the value and threats of deepfake content generated by artificial intelligence. Just under 40 percent view the technologies as positive and just under 50 percent view them negatively.
While such ambivalence would normally indicate the marketing for this technology has yet to reach a broader audience, the study also found that most consumers have already experienced deep fake technology in some form over various media outlets.
The study was sponsored by PinDrop and Voicebot. Pindrop makes tools for detecting voice clones, an audio deepfake, and will soon be branching out into video fakes. Voicebot is a media company producing newsletters and podcasts about the AI industry. it was conducted online during July of 2023 and completed by 2,027 U.S. adults age 18 or older.
Video vs audio
The study called deepfakes a broader term relating to both video and audio fakes in text, image, audio, or video formats replicating something that is real. Voice clones are a subset replicating the voice of a specific person.
PinDrop CEO Vijay Balasubramaniyan said the bigger problem at the moment is voice cloning because the amount of time and resources it takes to fake someone’s voice is much lower than video fakes. “Back in 2017 it took hours of a person’s voice to create a deep fake audio, but today you only need a three-second sample to make something that was 95 percent accurate.”
In contrast, even the best video deepfakes have several “tells” according to Balasubramaniyan. Eye movements, unnatural head movements, skin tones, and a general plastic look make deepfake videos fairly obvious, he said, “if you are paying attention.”
Concern and acceptance
The study said that more than half of U.S. adults are aware of deepfakes and voice clones, Consumers who are aware of deepfakes and voice clones express a great deal of concern with about 60% in both categories saying they were either “very” or “extremely” concerned and over 90% expressing some concern. At the same time, a significant number of consumers have dabbled with making their own deepfakes.
On the more realistic side, the study pointed out that consumers are concerned about applications involving sensitive personal identifiable information (PII) at risk and where false public information could have negative consequences. Specifically in banking, healthcare, and insurance, followed by government and media
“While consumers express significant concern about deepfakes and voice clones, there is a high degree of uncertainty about the readiness of leading institutions, the study explained. “Around 25% of U.S. consumers say they are “Neutral or Unsure” that businesses are ready for nefarious uses of the technology across key industries.”
Social media vs news media
However, the respondents believe that banks, insurance, and healthcare” take steps to protect them against risks. the gave a low-confidence rating for news and social media. “This is significant as these are channels driving the highest number of encounters.”
The top four categories where consumers have experienced deepfake video and audio content is YouTube, TikTok, Instagram, and Facebook, with an emphasis on video content. Movies, the news media, and television followed closely behind Facebook and Instagram. Consumers are likely to encounter voice clones on audio channels such as Spotify and phone calls. They were also significantly more likely to have created their own voice clone.
(This reporter recently received a phone call from celebrity Ed McMahon telling him he has won the Publishers Clearinghouse Contest. If McMahon hadn’t died in 2009 the conversation might have gone longer.)
The report concluded “There is a perfect storm of supply and demand for deepfakes and voice clones due to a convergence of rising quality and ease of access combined with mass distribution through social and traditional media. Deepfakes and voice clones may have introduced the “best of times” for creative pursuits. However, if you are focused on fraud prevention, copyright monitoring, or just want to protect your reputation, are you ready for “the worst of times?” The positive applications will take care of themselves. The real question is how to identify and mitigate the risks posed by negative uses.”
Lou Covey is the Chief Editor for Cyber Protection Magazine. In 50 years as a journalist he covered American politics, education, religious history, women’s fashion, music, marketing technology, renewable energy, semiconductors, avionics. He is currently focused on cybersecurity and artificial intelligence. He published a book on renewable energy policy in 2020 and is writing a second one on technology aptitude. He hosts the Crucial Tech podcast.
Pingback: Before you post, would you bet your job on it? - Cyber Protection Magazine