What is Synthetic Content?
The Federal Bureau of Investigation defines synthetic content as a broad spectrum of generated or manipulated digital content that includes images, video, audio, and text. In other words, it is content that has been automatically created.
Advances in AI (artificial intelligence) and ML (machine learning) have led to the creation of programmed content that seems to be real or manually created. For instance, each time the website ThisPersonDoesNotExist is refreshed, a very realistic image of a person who does not exist is generated by AI!
Generally, this content is considered as protected speech under the First Amendment. The FBI recently issued a notification that it anticipates malicious synthetic content attributed to foreign actors or cybercriminals which may result in an investigation.
Deepfake Technology
Deepfakes are videos or audio recordings that are digitally altered by AI. Deepfakes are becoming more difficult to detect. It is expected that high quality videos will soon be inexpensively created using downloaded software and apps by individuals with minimal technical knowledge. Combined with the virality of the internet, believable fake videos have the potential to compromise privacy, harm corporations, spread societal and political discord, and propagate disinformation. (Disinformation is false information that is meant to mislead; whereas misinformation is false information provided without malice.)
Deepfake video and audio can be used maliciously in emails, texts, and phone calls to convince recipients that the information is real. Reportedly in 2020, a Hong Kong bank manager authorized $35 million in transfers based on a call he believed to be from the director of a company with whom he’d previously spoken. It was a fraudster using AI to spoof the director’s voice.
Beneficial Uses
Deepfake technology also has a positive side. It is used as a tool to enhance communications during training, marketing, news reporting, and video chat bots. Some additional constructive uses include:
- Education: bring historical figures to life
- Language Translation: inexpensively generate high-quality videos in multiple languages to disseminate advertisements or corporate messages
- Healthcare: voice replacement to emulate the voice of a person who can no longer speak
- Entertainment: content creation, editing without a re-shoot; gaming technology
- Increase Revenue: enable people to authorize use of their images and get paid for “appearances” without traveling
Reuters has collaborated with Synthesia, a UK-based startup, to create automated personalized news reports for viewers. EY, formerly known as Ernst & Young, is exploring Synthesia’s technology for AI-created avatars to send videos instead of emails to clients. Refer this article to learn 15 ways to spot deepfake videos.
Avatars will become more lifelike with technology advances. Incidentally, Swedish pop group ABBA is building a special venue where members will be performing digitally via avatars with a live band in 2022.
Potentially Dangerous Repercussions
Deepfake technology will transform audio and video productions, but the ethical and legal considerations are immense. Recently, the documentary “Roadrunner: A Film About Anthony Bourdain” is based on remembering chef Anthony Bourdain and includes his synthetic audio, which was not initially disclosed and has triggered questions about ethical boundaries.
Deepfakes can be a threat to anyone depicted in such videos or audio content. Conversely, authentic content can be labelled as deepfakes to propagate distrust, create confusion and lead people to question reality. Some examples of potential risks include:
- Biometric Spoofing – for example, trick family members via vishing to send funds, authorize access to sensitive information, or distribute funds to hackers. Pay close attention to phone calls or voice messages that ask for funds.
- Bullying, Harassment
- Adult Content Videos
- Spear Phishing Attacks
- Impersonate Executives to get employees to commit fraud
- Fake Promotional Material
- Attempt by Competitors to Damage Reputation/Negatively Impact Shares of Public Companies
- Manipulate Money Markets or Stocks
- Re-frame History
- Election and Evidence Tampering
- Conspiracies by Foreign Adversaries
- Distort Emergency Alert Warnings/Public Service Announcements
- News Reporting of Disinformation
- Incite Violence
Future Safeguards
Companies continue to develop technologies to identify deepfakes. In June 2010 the best algorithm in Facebook’s Deepfake Detection Challenge accurately determined if a video was real or fake with 65% accuracy. In September 2021 M12, Microsoft’s venture capital fund, invested in deepfake verification startup Truepic.
In August 2021 the U.S. Senate Committee on Homeland Security and Governmental affairs voted unanimously to advance the bi-partisan Deepfake Task Force Act, which would establish a team with representatives from the Federal Government, higher education, and private or nonprofit organizations to investigate policy and technology strategies for limiting the damage of deepfake technology. The DEEPFAKES Accountability Act was introduced in June 2019.
