Difference between revisions of "Deepfake Misinformation"
| Line 1: | Line 1: | ||
| − | Deepfake is a form of media that has been altered by machine learning and artificial intelligence. It is often represented as videos or images that have been manipulated to make it appear as if a person is saying or doing something that they did not in reality actually say or do. The modified media produced can be misleading as the generated visual and audio content is not based on the original material, but rather fake content. <ref name="dave>Johnson, D. (n.d.). What is a deepfake? everything you need to know about the AI-powered fake media. Business Insider. Retrieved January 27, 2023, from https://www.businessinsider.com/guides/tech/what-is-deepfake </ref> Within recent years, the technology behind deepfakes has advanced rapidly, making it more accessible to create realistic fake videos and images. This growth of the technology has led to many concerns about the potential of deepfakes being used to spread misinformation and propaganda and also includes harassment and blackmailing. | + | Deepfake is a form of media that has been altered by machine learning and artificial intelligence. It is often represented as videos or images that have been manipulated to make it appear as if a person is saying or doing something that they did not in reality actually say or do. The modified media produced can be misleading as the generated visual and audio content is not based on the original material, but rather fake content. <ref name="dave">Johnson, D. (n.d.). What is a deepfake? everything you need to know about the AI-powered fake media. Business Insider. Retrieved January 27, 2023, from https://www.businessinsider.com/guides/tech/what-is-deepfake </ref> Within recent years, the technology behind deepfakes has advanced rapidly, making it more accessible to create realistic fake videos and images. This growth of the technology has led to many concerns about the potential of deepfakes being used to spread misinformation and propaganda and also includes harassment and blackmailing. |
| + | |||
| + | ==History of Deepfake and How it Works== | ||
| + | ===History=== | ||
| + | The idea of photo manipulation started in the 19th century <ref name="arnold">Arnold. (2021, January 9). Deepfake history: When was deepfake technology invented? Deepfake Now. Retrieved January 27, 2023, from https://deepfakenow.com/deepfake-history-when-invented/ </ref>, and the general idea of media manipulation started to rise as technology advanced which included video formats. However, using computer algorithms to manipulate media is a newer idea that started in 1997 with the video rewrite program. <ref name="arnold" /> This program had the ability to put false audio over a video which allowed the manipulator to make a person in the video say anything. However there would be only a small amount of academic research done on this subject up until 2017 which would be the year where the term “deepfake” was used to describe the altered media content. | ||
| + | ===How it Works=== | ||
| + | The most common technique to create deepfake software is to use machine learning algorithms to generate the altered content based on a large amount of data input. Primarily these techniques are based on autoencoders and on generative adversarial networks (GAN). <ref name="recfaces">What is deepfake?all about Deepfake Technology. RecFaces. (2023, January 19). Retrieved January 27, 2023, from https://recfaces.com/articles/what-is-deepfake#2 </ref> Autoencoders are a group of self-supervised neural networks that learn to copy their own input. An autoencoder consists of 3 components: an encoder, a code, and a decoder. The encoder compresses the input data and produces the code after the decoder reconstructs the input based only on the code. <ref name="recfaces" /> Another technique that is often used through deepfake is the generative adversarial networks (GAN), which also learns from input data in order to generate new data. The system is trained by two distinctive neural networks: a generator and a discriminator where the generator discovers regularities or patterns in the input dataset and learns to reproduce them. This generated data is used by the discriminator to compare to the real data and check if it can distinguish between the two. This system is trained until the discriminator no longer confuses the generated data with the real data <ref name="recfaces" />. | ||
| + | |||
| + | ==How Deepfake is Used to Spread Misinformation== | ||
| + | ===Political Public Figures=== | ||
| + | Deepfake is uniquely effective at spreading misinformation especially within the setting of democratic discourse. Deepfake videos depicting public figures making incendiary comments or behaving inappropriately may alter election outcomes. In addition, on the reverse, as deepfakes become more well known, public officials caught on camera can also exploit this and claim that a real video is a deepfake. Without clear methods to distinguish what is real from what is not real, the public may lose trust in the media and other public institutions. <ref name="regulatory">The Regulatory Review. (2021, August 23). Responding to deepfakes and disinformation. The Regulatory Review. Retrieved January 27, 2023, from https://www.theregreview.org/2021/08/14/saturday-seminar-responding-deepfakes-disinformation/ </ref> | ||
| + | |||
| + | ===Pornography=== | ||
| + | Women have been the largest target for deepfake use. Sensity AI, a research company that has tracked online deepfake videos since December of 2018, has consistently found that between 90% and 95% of them are nonconsensual porn and about 90% of that is nonconsensual porn of only women. <ref name="karen">Hao, K. (2021, February 16). Deepfake porn is ruining women's lives. now the law may finally ban it. MIT Technology Review. Retrieved January 27, 2023, from https://www.technologyreview.com/2021/02/12/1018222/deepfake-revenge-porn-coming-ban/ </ref> In large part, these are released as revenge porn—real intimate photos released without consent. This is particularly harmful as the photos and videos are not only released to the public, but the actual content is also inaccurate. | ||
| + | |||
| + | ===General News=== | ||
| + | Another concern is that deepfakes could be used to create fake news stories or other types of misinformation that are spread through social media and other online platforms. These can be used to spread rumors or false information about individuals or groups, potentially causing harm to the reputation of those involved. In addition, this could lead to confusion and mistrust among the public, as well as potentially dangerous consequences such as stock market fluctuations or even violence. Conspiracy theories are also potentially using deepfake to provide proof, which can be dangerous to the public's understanding of general events. | ||
| + | |||
| + | ===Ethical Concerns=== | ||
| + | Disinformation and Propaganda | ||
| + | The misuse of deepfake can be detrimental as it can be used to spread propaganda to groups of individuals. This could lead to confusion and mistrust among the public, and could even be used to influence elections or sway public opinion on important issues. Additionally, deepfakes can enable the least democratic and authoritarian leaders to thrive as they can leverage the ‘liar’s dividend’, where any inconvenient truth is quickly discounted as ‘fake news’. <ref name="ashish">Jaiman, A. (n.d.). Debating the ethics of deepfakes. ORF. Retrieved January 27, 2023, from https://www.orfonline.org/expert-speak/debating-the-ethics-of-deepfakes/ </ref> Deepfakes and synthetic media may have a large impact on the outcome of elections. This disinformation creates large amounts of harm to individuals because it prevents their ability to make informed decisions in their own best interests. Intentionally distributing false information about the opposition or presenting an alternate truth for a candidate in an election manipulates voters into serving the interests of the deceiver. <ref name="ashish" /> The ethics of deepfake is questioned as they can be used with the intent to deceive, intimidate, and inflict reputational harm to extend disinformation. | ||
| + | |||
| + | Blackmail and Revenge Porn | ||
| + | Deepfake has the potential to be used as blackmail in situations where individuals can threaten to release fake content in hopes of gaining money or other motives. In addition, deepfake porn can be also used to blackmail people (mostly females) as they don’t want to lose their reputation and image. The major issue can be attributed to mimicking a person’s likeness without their permission and in the case of deepfake porn, this can be considered unethical since the victim of the deepfake is being used as a source of pleasure and entertainment, without consent. <ref name="kiara"> Goodwine, K. (2022, December 14). Ethical considerations of Deepfakes. Prindle Institute. Retrieved January 27, 2023, from https://www.prindleinstitute.org/2020/12/ethical-considerations-of-deepfakes/ </ref> Additionally, the use of non consensual deepfake can be particularly harmful if someone’s likeness is used without their knowledge. | ||
| + | |||
| + | == References == | ||
Revision as of 07:15, 27 January 2023
Deepfake is a form of media that has been altered by machine learning and artificial intelligence. It is often represented as videos or images that have been manipulated to make it appear as if a person is saying or doing something that they did not in reality actually say or do. The modified media produced can be misleading as the generated visual and audio content is not based on the original material, but rather fake content. [1] Within recent years, the technology behind deepfakes has advanced rapidly, making it more accessible to create realistic fake videos and images. This growth of the technology has led to many concerns about the potential of deepfakes being used to spread misinformation and propaganda and also includes harassment and blackmailing.
Contents
History of Deepfake and How it Works
History
The idea of photo manipulation started in the 19th century [2], and the general idea of media manipulation started to rise as technology advanced which included video formats. However, using computer algorithms to manipulate media is a newer idea that started in 1997 with the video rewrite program. [2] This program had the ability to put false audio over a video which allowed the manipulator to make a person in the video say anything. However there would be only a small amount of academic research done on this subject up until 2017 which would be the year where the term “deepfake” was used to describe the altered media content.
How it Works
The most common technique to create deepfake software is to use machine learning algorithms to generate the altered content based on a large amount of data input. Primarily these techniques are based on autoencoders and on generative adversarial networks (GAN). [3] Autoencoders are a group of self-supervised neural networks that learn to copy their own input. An autoencoder consists of 3 components: an encoder, a code, and a decoder. The encoder compresses the input data and produces the code after the decoder reconstructs the input based only on the code. [3] Another technique that is often used through deepfake is the generative adversarial networks (GAN), which also learns from input data in order to generate new data. The system is trained by two distinctive neural networks: a generator and a discriminator where the generator discovers regularities or patterns in the input dataset and learns to reproduce them. This generated data is used by the discriminator to compare to the real data and check if it can distinguish between the two. This system is trained until the discriminator no longer confuses the generated data with the real data [3].
How Deepfake is Used to Spread Misinformation
Political Public Figures
Deepfake is uniquely effective at spreading misinformation especially within the setting of democratic discourse. Deepfake videos depicting public figures making incendiary comments or behaving inappropriately may alter election outcomes. In addition, on the reverse, as deepfakes become more well known, public officials caught on camera can also exploit this and claim that a real video is a deepfake. Without clear methods to distinguish what is real from what is not real, the public may lose trust in the media and other public institutions. [4]
Pornography
Women have been the largest target for deepfake use. Sensity AI, a research company that has tracked online deepfake videos since December of 2018, has consistently found that between 90% and 95% of them are nonconsensual porn and about 90% of that is nonconsensual porn of only women. [5] In large part, these are released as revenge porn—real intimate photos released without consent. This is particularly harmful as the photos and videos are not only released to the public, but the actual content is also inaccurate.
General News
Another concern is that deepfakes could be used to create fake news stories or other types of misinformation that are spread through social media and other online platforms. These can be used to spread rumors or false information about individuals or groups, potentially causing harm to the reputation of those involved. In addition, this could lead to confusion and mistrust among the public, as well as potentially dangerous consequences such as stock market fluctuations or even violence. Conspiracy theories are also potentially using deepfake to provide proof, which can be dangerous to the public's understanding of general events.
Ethical Concerns
Disinformation and Propaganda The misuse of deepfake can be detrimental as it can be used to spread propaganda to groups of individuals. This could lead to confusion and mistrust among the public, and could even be used to influence elections or sway public opinion on important issues. Additionally, deepfakes can enable the least democratic and authoritarian leaders to thrive as they can leverage the ‘liar’s dividend’, where any inconvenient truth is quickly discounted as ‘fake news’. [6] Deepfakes and synthetic media may have a large impact on the outcome of elections. This disinformation creates large amounts of harm to individuals because it prevents their ability to make informed decisions in their own best interests. Intentionally distributing false information about the opposition or presenting an alternate truth for a candidate in an election manipulates voters into serving the interests of the deceiver. [6] The ethics of deepfake is questioned as they can be used with the intent to deceive, intimidate, and inflict reputational harm to extend disinformation.
Blackmail and Revenge Porn Deepfake has the potential to be used as blackmail in situations where individuals can threaten to release fake content in hopes of gaining money or other motives. In addition, deepfake porn can be also used to blackmail people (mostly females) as they don’t want to lose their reputation and image. The major issue can be attributed to mimicking a person’s likeness without their permission and in the case of deepfake porn, this can be considered unethical since the victim of the deepfake is being used as a source of pleasure and entertainment, without consent. [7] Additionally, the use of non consensual deepfake can be particularly harmful if someone’s likeness is used without their knowledge.
References
- ↑ Johnson, D. (n.d.). What is a deepfake? everything you need to know about the AI-powered fake media. Business Insider. Retrieved January 27, 2023, from https://www.businessinsider.com/guides/tech/what-is-deepfake
- ↑ 2.0 2.1 Arnold. (2021, January 9). Deepfake history: When was deepfake technology invented? Deepfake Now. Retrieved January 27, 2023, from https://deepfakenow.com/deepfake-history-when-invented/
- ↑ 3.0 3.1 3.2 What is deepfake?all about Deepfake Technology. RecFaces. (2023, January 19). Retrieved January 27, 2023, from https://recfaces.com/articles/what-is-deepfake#2
- ↑ The Regulatory Review. (2021, August 23). Responding to deepfakes and disinformation. The Regulatory Review. Retrieved January 27, 2023, from https://www.theregreview.org/2021/08/14/saturday-seminar-responding-deepfakes-disinformation/
- ↑ Hao, K. (2021, February 16). Deepfake porn is ruining women's lives. now the law may finally ban it. MIT Technology Review. Retrieved January 27, 2023, from https://www.technologyreview.com/2021/02/12/1018222/deepfake-revenge-porn-coming-ban/
- ↑ 6.0 6.1 Jaiman, A. (n.d.). Debating the ethics of deepfakes. ORF. Retrieved January 27, 2023, from https://www.orfonline.org/expert-speak/debating-the-ethics-of-deepfakes/
- ↑ Goodwine, K. (2022, December 14). Ethical considerations of Deepfakes. Prindle Institute. Retrieved January 27, 2023, from https://www.prindleinstitute.org/2020/12/ethical-considerations-of-deepfakes/
