Dawn of the Deep Fake : The Need for Government Involvement
DeepFake Dangers and Origins
Since the creation of artificial intelligence in the mid 20th century, technology has been improving at an unparalleled pace. Relatively recently, a new form of technology emerged called Deepfakes — an innovation that replaced or synthesized someone’s likeness onto another video, picture, or recording. Deepfakes began to grow in scale, catching the attention of the US Defense Advanced Research Projects Agency, which later took on projects to investigate the efficacy of these AI systems and to optimize their capabilities. As the software to create Deep Fakes becomes more widely available to amateurs and the general public, it is increasingly urgent to implement better ways to regulate this novel technology from potential abuses.
The spread of Deepfakes has been largely facilitated through their multimodal nature. According to research done by renowned Oxford Computer researcher Dr. Floridi, Deepfakes can include numerous types of media, examples including a synthesized speech by John F. Kennedy and AI generated paintings by Rembrandt. This illustrates the numerous types of Deepfakes and the variety of different kinds of platforms Deep Fakes are accessible and can be created on. This facilitates their spread and the spread and their prevalence on the internet.
The legislation being passed to regulate these controversial AI advancements is minimal and concern by the public is significant. In a 316 participant study done by Dr. Cochran and Dr. Napshrin, the public does not have high confidence in the ability of technology to combat the issues of Deepfakes, however, they do believe that Deep Fakes are an important issue that platforms must account for. With the increasing quality of Deepfakes, more concerns are arising among the public that the spread of misinformation by hyperrealistic Deep Fakes is an issue that must be addressed.
Effect of DeepFakes
While many of these Deep Fake videos seem harmless, “96% of the total deepfake videos found online…” are nonconsensual pornagraphy according to a Canadian Review on Deepfake policy, and “by September 2019 the top four dedicated deepfake pornography websites alone had garnered >134 million views.” The issue of Deepfakes is a human rights one and can cause people to lose credit to their own identity due to lack of laws. According to a study done by Eriq Gardner, a senior editor at the Hollywood Reporter, people who create or superimpose images to create Deepfakes are considered creators and get the ultimate credit to Deepfakes according to current copyright laws, causing people in the videos to lose credit to their own identities. With the high quality of current Deepfakes, as well as their synthetic nature, it is getting easier to create Deepfakes and to take away credit from the people whose images are used in the Deep Fakes.
Although Deepfake pornagraphy is the main type of circulated deepfakes, another widely distributed type of Deepfake is those aimed at the general public that aid in the spread of disinformation by creating vetted videos of celebrities and elected officials. In addition to creating false narratives that can hurt their reputations through auditory and visual manipulation of their images, these Deepfakes can have major impacts on the outcomes of elections.
Research done on the effects of Deepfakes during US elections shows that Deepfakes have helped in elevating fears and in strengthening negative emotions towards candidates during elections, thus influencing voter behaviour. Since voter behaviour and their perception of current events leads to their decisions on who and what to vote for during elections, Deepfakes can ultimately have lasting impacts on elections and US society.
Further research done by Assistant professor Ahmed at Nanyang Technological University for political communication, shows that people who are more politically active are more likely to spread Deepfakes. With this research, it can be argued that since politically people are more likely to vote and more likely to have been influenced by government-targeted Deepfakes, there is a large chance that Deepfakes of politicians and government officials can have major impacts on elections and who people vote for.
Furthermore, Deepfakes can have enormous impacts on human rights by undermining the faith of the general public in the media. A study done by Sam Gregory, a Harvard graduate of Public policy, illustrates how the presence of Deepfakes hinders reporting of human trafficking and human rights violations. With increasing numbers of doctored videos, it is harder to tell truth from fiction, making it harder to compile data on human rights violations. As the public becomes more aware of the misinformation being spread by Deepfakes, their trust gets diminished. With less trust in the media, victims of human rights violations are less likely to be believed due to the copious amounts of fake documentations of human rights violations on the internet. This study can be translated to not only human rights violations but spread of any sort of integral information concerning public health, elections, and other violations.
As an evolving conversation, Deepfake technology is only evolving and new advancements in AI render even highly sophisticated facial recognition technologies unable to detect a Deepfake from an actual image. The technology is easily accessible by the public granting people with increasing avenues of accessing and creating Deep Fakes, warranting an increased need in regulation of these Deepfakes.
What Must Be Done?
Due to the international and dangerous nature of Deepfakes, as well as human rights violations and election sabotage dangers they pose, the US government must take measures to regulate and prevent Deepfakes. Since this is a relatively new technology and an evolving conversation, there is little to no legislation helping to curb the spread of Deepfakes. Increased regulations and more specific rhetoric in laws concerning Deepfakes need to be implemented to prevent their spread of false information and defamation of elected officials, in conjunction to more funding for research on high quality Deepfakes.
Although research might be seen as the way to proceed in combating Deep fakes, the technology for Deepfakes is constantly improving, thus constant research to combat these improvements must be done. Laws and preventive measures are needed to help curb the spread of Deepfakes since research can take years to complete. Current legislation is too lenient on checking for misinformation, highlighted by Communications Decency Act of 1996, which allows many sites to publish information without checking for misleading or erroneous information, most notably Facebook and Google. These sites must face heavier scrutiny and have more regulation by the government to prevent the spread of misinformation.
A definition for Deepfake must be established in law, as well as copyright laws giving rights of Deep Faked work to the people targeted by Deepfakes, especially in instances where it is used against their permission. Ultimately, drastic change is needed to combat the growing dangers of Deepfakes and more Deepfake-specific government legislation should be the first, if not most crucial step.