Last year, Mark Zuckerberg, CEO of Facebook’s video went viral on Instagram in which he claimed that he has control over billions of people’s data, their lives and the future as well as know their secrets. It was not an indented effort of spreading misinformation instead was just an art project. The company did not remove it from the platform, saying it does not violate any rules of the platform.
Some of the people took as a failure of giant tech companies as they are not completely ready to deal with the deepfakes – AI-generated fake media content. It is not only Facebook, deepfakes in general are not easy to moderate. Though they can be identified through some tricks but due to the broad category, in an attempt to compress it, the safe and harmless AI-generated content can also be marginalized.
If deepfakes are banned, it means that the popular filters on various social media platforms, like gender swap on Snapchat or the aged filter, will also be removed. In case politically misleading deepfakes are banned, it will lead to moderation issues that tech companies have been facing from the last few years. There is no simple algorithm that can detect AI-edited content automatically and banning it means there will be more workload and dependability on human moderators. Companies like Facebook may have much difficulty in handling data to an extensive level.
Facebook and Reddit recently announced their moderation policies which included deepfakes to some extent however it was not dealt on a broad level.
Facebook said it will not allow any misleading or manipulative content that is edited or synthesized using artificial intelligence. Content that might not be easy to recognize by common people and they mistakenly take it in the wrong context will be removed from the platform. Policy set for the content created using AI or machine learning does not cater to the parody or satire content neither misleading edits that are made using the traditional means.
Reddit did not directly mention AI rather only talked about removing content that impersonates anyone else or is presented in a misleading or deceptive way. The platform however excluded the satire or parody and rather focused on the context of the content.
Many people identified loopholes in these policies. There are some obvious missing issues that needed to be cater, making it difficult to differentiate between deepfakes and the moderation issues on the platform. In recent times, many of the deepfake issues have been dealt with like political misinformation and with time the problem is expected to grow worse.
Formerly it was not easy to use deepfake software and required some technical skills however with advancement, making deepfakes is going to be a lot easier. Apps used to edit video through AI are now common that are used to create jokes, for brand tweets and even harassment and bullying.
Recently an app with the name Doublicat was launched on app stores that replaces the face on famous GIFs. According to the CEO of the company, the app would soon allow replacing face on any content, which will be possible with the use of a new machine learning methods, GAN.
Though results are impressive but it is easier to see that it is not real and is rather modified.
TikTok and Snapchat have also experimented with deepfake features however TikTok’s owner company ByteDance does not plan to introduce it on the platform.
It is also expected that with time people will become accustomed to what technology is capable of and they will be more conscious and recognize the underlying issues and propaganda.
Though Facebook and Reddit somehow made their policies that included deepfake issues but it showed that the main problem is still not addressed. How and to what extent deepfakes will be allowed and who will decide whether it is a lie or not.
Deepfakes are a threat to the online truth and will raise more questions than ever before.
Photo: Shutterstock / Lightspring
Read next: This DataViz Shows Who Owns the Most Visited News Sites in the World
Some of the people took as a failure of giant tech companies as they are not completely ready to deal with the deepfakes – AI-generated fake media content. It is not only Facebook, deepfakes in general are not easy to moderate. Though they can be identified through some tricks but due to the broad category, in an attempt to compress it, the safe and harmless AI-generated content can also be marginalized.
If deepfakes are banned, it means that the popular filters on various social media platforms, like gender swap on Snapchat or the aged filter, will also be removed. In case politically misleading deepfakes are banned, it will lead to moderation issues that tech companies have been facing from the last few years. There is no simple algorithm that can detect AI-edited content automatically and banning it means there will be more workload and dependability on human moderators. Companies like Facebook may have much difficulty in handling data to an extensive level.
Facebook and Reddit recently announced their moderation policies which included deepfakes to some extent however it was not dealt on a broad level.
Facebook said it will not allow any misleading or manipulative content that is edited or synthesized using artificial intelligence. Content that might not be easy to recognize by common people and they mistakenly take it in the wrong context will be removed from the platform. Policy set for the content created using AI or machine learning does not cater to the parody or satire content neither misleading edits that are made using the traditional means.
Reddit did not directly mention AI rather only talked about removing content that impersonates anyone else or is presented in a misleading or deceptive way. The platform however excluded the satire or parody and rather focused on the context of the content.
Many people identified loopholes in these policies. There are some obvious missing issues that needed to be cater, making it difficult to differentiate between deepfakes and the moderation issues on the platform. In recent times, many of the deepfake issues have been dealt with like political misinformation and with time the problem is expected to grow worse.
Formerly it was not easy to use deepfake software and required some technical skills however with advancement, making deepfakes is going to be a lot easier. Apps used to edit video through AI are now common that are used to create jokes, for brand tweets and even harassment and bullying.
Recently an app with the name Doublicat was launched on app stores that replaces the face on famous GIFs. According to the CEO of the company, the app would soon allow replacing face on any content, which will be possible with the use of a new machine learning methods, GAN.
Though results are impressive but it is easier to see that it is not real and is rather modified.
TikTok and Snapchat have also experimented with deepfake features however TikTok’s owner company ByteDance does not plan to introduce it on the platform.
It is also expected that with time people will become accustomed to what technology is capable of and they will be more conscious and recognize the underlying issues and propaganda.
Though Facebook and Reddit somehow made their policies that included deepfake issues but it showed that the main problem is still not addressed. How and to what extent deepfakes will be allowed and who will decide whether it is a lie or not.
Deepfakes are a threat to the online truth and will raise more questions than ever before.
Photo: Shutterstock / Lightspring
Read next: This DataViz Shows Who Owns the Most Visited News Sites in the World
