With the ever-increasing popularity of Deepfake technologies, the pressure on governments and social media titans to regulate synthetic content, media with manipulated audio or video data, also increases. The impact of those regulations varies significantly from country to country. However, as Machine Learning (ML) and Deepfake technologies change and improve, so will the laws and regulations.
East and West, the two extremes of regulation
On the east side of the globe, Beijing is actively attempting to control the spread and utilization of Deepfake content. According to new policies, Chinese social media platforms are prohibited from suggesting manipulated media via the recommendations tab. More specifically, Deepfake content is not to be available in the “Read next” or similar links available on the timeline. It is imperative to mention that this is not the first regulation pushed by the Chinese government regarding Deepfake media, as an earlier prohibition has already banned news providers from sharing information created using ML and Deepfake technologies without labeling it as such.
On the opposite end, the US government has been considerably less enthusiastic about presenting significant laws or regulations. Deepfake content and, generally, any media is protected by the First Amendment, prohibiting the government from censoring or regulating free speech. Currently, the only regulations even close to restraining Deepfake content are defamation laws, providing minimal protection against online harassment, particularly in the form of altered content of sexual nature. Some bills have been presented at State level, like the 2019 bill outlawing the creation of political Deepfakes in Texas.
Deepfake and social media
Western tech giants have not hesitated in taking action against false media. Major social networks like Facebook and Twitter have introduced new regulations and updated their policies, requiring posts and tweets that can potentially spread confusion or incorrect information to be labeled as altered media. Posts can also get outright removed if malicious or intimidating, possibly resulting in account bans. Moreover, companies are actively investing in technologies capable of detecting Deepfakes. In 2020 Facebook held a contest to find software capable of detecting altered content, with the winning program having a 66% success rate.
The delicate balance
So, why not just completely ban Deepfake media from all social platforms? Besides the apparent conflict with free speech, it is essential to realize that not all deepfake content is malicious or harmful in nature. Deepfake technologies have various practical applications, ranging from creating engaging video material for history classes to streamlining expensive processes like CGI and video editing. Deep learning technologies are seeing new uses in the medical field, with tumor spotting ML algorithms using Deepfake generated ‘fake’ MRI data for training.
While careful government and social media regulation are imperative, media literacy could be the saving grace to dealing with antagonistic and predatory content. By informing regular users about the possibility of ill-natured content, it is possible to limit the spread of fake media and mitigate the adverse effects. Through a combination of government regulations, information campaigns, and social media responsibility, Deepfake technologies can get the positive reputation they deserve.