Below are the opinions of two experts from RMIT University on the harmful effects of deepfakes and proposed solutions to this problem:
 |
| Dr Jonathan Crellin (left) and Dr Nguyen Van Thang Long (right). (Source: RMIT University) |
Cybercriminals exploit new technology in unforeseen ways
Dr Jonathan Crellin, Head of Information Security, Faculty of Science, Engineering and Technology, RMIT University :
Deepfake is a machine-generated image that can combine images or videos from multiple sources, creating a very realistic image, video, or even audio.
Deepfakes rely on a technique in AI called machine learning, which can replace and integrate elements like a person's face into another image or video.
One example of deepfake use is combining pornographic content with photos of Taylor Swift to create fake images.
To do this, the mastermind needed some images so that the software could learn about the singer's facial expressions, then combine them with pornographic content to create obscene photos, causing the singer's image to be tarnished.
In fact, it is rumored that these photos were released by a Telegram group and were created using Microsoft Designer tool with integrated AI support.
Anyone can become a victim, as the bad guys only need a picture, video or audio recording of the person being impersonated.
Fake news in many forms has the potential to be used to create false news, and it is safe to assume that the upcoming US presidential election will be filled with such fake news.
Currently, lawmakers around the world are looking to enact laws against this type of imagery.
Some approaches are starting to be adopted in the United States in terms of legislation, such as relying on civil lawsuits, or laws to address “the dissemination of sexually explicit images of a person generated by AI without that person’s consent.”
China also introduced new regulations that would allow prosecution of those who spread AI-generated images.
The UK has made sharing deepfake content illegal under its online safety act.
How to detect or prevent this?
The first step is to reduce the number of photos, videos, or audio recordings you have online. Make sure you only share them with people you know and not post them widely online. Once something is on the Internet, it is almost impossible to remove it.
The second way is to agree on a secret word with your family to authenticate the call, reducing the risk of falling into the trap of a fake call.
Images, especially videos, can have strange (staged-looking) errors, if you spot these errors then there is a high chance that the image or sound has been faked.
Another technique that can be used is to search for “reverse image” on Google or other search engines, to determine the source of the original image.
The final lesson is don't blindly believe what you see, the camera (or AI) can lie!
Deepfakes pose a huge threat to celebrities and politicians
Dr. Nguyen Van Thang Long, Senior Lecturer, Faculty of Communication and Design, RMIT University :
With the proliferation of fake news from deepfakes, the media teams of celebrities and politicians need to have resources available to monitor and respond promptly to fake news or continuously correct false news.
If deepfakes are systematically combined with organized forms of “dirty PR”, this task becomes even more difficult due to the proliferation of conflicting information, and fake news and negative news will always be shared more than positive news.
Usually, when seeing news shared on social networks, individuals often have the habit of confirming information through official media channels.
With deepfake content flooding social media, it is increasingly difficult and time-consuming to verify the accuracy of mainstream news, requiring in-depth research and verification techniques.
The longer we delay verifying news and sources, the more likely false, fabricated or misleading information will spread due to the rapid speed of sharing and commenting on social media.
This exacerbates the underlying problem and has the potential to lead to social unrest, especially if the content relates to political speech, religion, gender, business strategy or macroeconomic issues.
In the context of deepfake proliferation, the most effective risk management strategy remains to maintain consistent communication channels, through popular social media platforms, websites, or face-to-face meetings, between businesses, celebrities, politicians and key stakeholders such as fans, press, communities, and employees.
 |
The increased use of AI brings both benefits and unexpected new problems. Illustration photo. (Source: Freepik) |
By maintaining these communication channels, receiving information related to deepfakes becomes faster, allowing for timely and effective rumor correction, exposing misinformation at the outset.
However, companies, celebrities, and politicians need to develop a crisis response plan specifically for deepfakes. For example, who is reporting, through which media channels, criteria for verifying information through evidence and reputable sources, establishing a timeline for resolving rumors, and outlining a strategy for restoring reputation.
With a well-prepared, methodical plan, handling a deepfake crisis will be more feasible, minimizing possible unfortunate consequences.
Source
Comment (0)