My purpose in this section is not to scare people but to give a more balanced perspective on the pros and cons of deep fake video and audio technologies. The key point is while we take advantage of these technologies, we also need to be mindful of their risks and take cautions proactively and collaboratively.
Ethics and law are the critical aspects of emerging technologies like artificial intelligence. Even though there are many concerns of AI, I only touch on the deep fake video and audio technologies in this post. I plan to post more on various aspects of ethical and legal matters related to AI.
Consumers love deep fake videos and audio documents for various reasons, such as education and entertainment purposes. However, entertaining and educative videos are usually received well by the public hence have the potential to go viral with the power of social media.
You probably have seen Barrack Obama’s face voiced by Jordan Peele in this video in 2018. Jordan, an actor, comedian, and filmmaker who produced the video using artificial intelligence to warn about the future of fake news.
Major concerns of deep fake technologies are related to ethics and laws. Fraud tops the list in the industry. For example, tampering with medical imagery is a critical concern in the healthcare industry.
In addition, considering the power of cryptocurrencies and social engineering, fake deep videos can be used for blackmailing public figures, wealthy people, and business organizations.
When these powerful techniques go to the wrong hands like terrorists, the impact can be significant. The fear is valid because the AI tools are available to the masses. Sophisticated and organized efforts using AI can even threaten our existence on this globe. I will touch on this critical point in another article.
Visual and audio features can be easily manipulated using AI. Speed adjustment is the most commonly used technique to create new patterns. These techniques can change the context of content easily. When AI enters the equation in video editing and distribution, it induces reasonable concerns and heighten fears.
I added an audio version of an article at the end. A reader sent me an interesting private message when she heard a female voice, by saying that “I thought you were a man from your profile picture and name.” While this message gave me giggles at the time, it also raised concerns as people can easily get confused with gender differences in audio files.
We use audio files for creating podcasts. A media company offered me a free trial to convert my blogs to podcasts. I am so impressed by the efficiency of turning my blog posts to podcasts with a few clicks. This is an excellent opportunity for bloggers who aspire to publish their text materials as podcasts.
But this opportunity also brings the risk of blog posts changing to podcasts with altered content beyond the control of bloggers. For example, someone can copy and paste the blog with minor edits can turn a blog post to a podcast. This simple use case might have profound implications, such as breaching copyright rules and causing a financial impact on bloggers and writers.
Significant concerns of deep fake videos and audios are fraud, damaging organizations, defaming people, eroding public trust in governments, weakening journalism, and dangering public safety. All of these points are possible, and none of them is a trivial matter.
Even though we are still at the nascent stage of technology, we expect mobile technologies to make these tools more accessible. For example, anyone soon can create deep fake videos and audio files using smartphone applications. With this mass usage, it can be too arduous to control the situation.
We know that freedom of expression is a double-edged sword and so is the current situation of deep fake technologies. While AI technologies bring many benefits to society, they also pose many critical risks for our safety, security, and quality of life.
Governments are slow in addressing the deep fake issues, but there are some promising acts from the state government in the US and the government of China. While some media companies are also slow in addressing the issues, at least Twitter and Facebook created some policies.
Twitter took some measures as communicated in this official blog titled “Building rules in public: Our approach to synthetic & manipulated media”. In summary, “You may not deceptively share synthetic or manipulated media that are likely to cause harm. In addition, we may label Tweets containing synthetic and manipulated media to help people understand the media’s authenticity and to provide additional context.” This tweet posted by Twitter in February 2020 received 1996 quotes that may give you ideas from various Twitter users.
Facebook developed some measures in 2020. For example, this artefact Enforcing Against Manipulated Media, posted by Monika Bickert, Vice President, Global Policy Management, pointed out that Facebook was strengthening policy toward misleading manipulated videos that have been identified as deep fake.
Facebook also partnered with industry leaders and academic experts to create the Deepfake Detection Challenge (DFDC) to accelerate the development of new ways to detect deep fake videos in 2020. “By creating and sharing a unique new dataset of more than 100,000 videos, the DFDC has enabled experts from around the world to come together, benchmark their deepfake detection models, try new approaches, and learn from each other.”
According to InfoSecurity Magazine, “YouTube became the first major social media platform to remove a deep fake video of Speaker of the United States House of Representatives Nancy Pelosi. The clip had been slowed down to make it appear that Pelosi was slurring her words, causing her to come across as almost drunk enough to start a conga line.”