In September 2019, we released our State of Deepfakes report, the first research of its kind providing data-driven analysis of the deepfake landscape. Since we collected this report’s data in July 2019, the technologies and activity surrounding deepfakes have continued to rapidly evolve. This blog post outlines some of our continued data analysis tracking deepfakes’ evolution as of June 2020.
Number of deepfakes identified online
Since December 2018, the number of deepfakes online is roughly doubling every six months, confirming a continued exponential growth. As disclosed in The State of Deepfakes 2019, the total number of deepfakes had reached 14,678 in July 2019, representing an almost 100% increase since our previous measurement of 7,964 videos in December 2018. Since then, we have continued to observe a significant growth in the number of deepfakes videos online, as well as an increasing number of deepfake sources and creators. As of June 2020, our monitoring activities identified 49,081 deepfake videos, an increase of more than 330% since July 2019.
Do you want get access to the world largest database of detected deepfakes?
Most targeted sectors
As well as monitoring the overall growth in the number of deepfake videos online, our analytics also capture key information on the individuals who are targeted in these videos. Our analysis from July 2019 found the vast majority of targets (over 95%) worked in the entertainment industry, with a remaining small percentage of targets coming from business, news media, and politics. As of June 2020, entertainment is still the most targeted industry at 62.7%, which combined with the newly separated categories of fashion (21.7%) and sport (4.4%) totals 88.9% of all targets. Of the 62.7% of targets from the entertainment industry, we identified a notable increase in the number of Instagram, Twitch, and Youtube personalities being targeted. We also identified an increase in targets from business (4.1%) and politics (4%) backgrounds.
Most targeted countries
Our July 2019 research found individuals from a wide range of nationalities were targeted in the deepfake videos we identified online. The majority of these videos featured targets from western countries (USA and UK) with a notable number of videos also featuring South Korean and Indian targets. Our June 2020 analysis presents a similar picture of targets’ nationality, with USA and UK nationals making up a slightly lower (61%) majority of the total unique individuals targeted. South Korean, Indian, and Japanese nationals also make up a significant proportion of targets, a finding that correlates to our research observations that deepfake activity continues to grow in Asia.