Jakarta, CNBC Indonesia – As technology becomes more sophisticated today, Deepfakes or AI-generated content that is difficult to distinguish from real photos, videos and sounds are starting to be detected on Instagram, Facebook and Threads. Therefore, Meta took steps by providing special labels and signs.
Meta President of Global Affairs Nick Clegg stated that policies for monitoring and labeling AI-generated content will be implemented, for content created using the Meta platform and platforms outside of Meta.
AI-generated content can be recognized by an invisible “stamp” embedded in the image file. If the stamp is detected, Meta will provide a special label on content uploaded on Facebook, Instagram and Threads.
Currently, AI monitoring policies and features are in place for content created using Meta technology. In the next few months, the same system will be applied to content created by platforms owned by OpenAI, Midjourney, Shutterstock and Google.
According to Reuters, Meta's steps indicate the direction of social media and internet platform policies to mitigate the negative impacts of circulating fake AI-engineered content. Similar methods have been implemented in the last 10 years to remove prohibited content such as content depicting violence and exploitation of children.
Clegg is confident that Meta's system can recognize every AI-generated image uploaded to its platform. However, the company founded by Mark Zuckerberg is still perfecting a similar system for recognizing AI-generated video and audio.
“Even though this technology is still new, especially for audio and video, we hope it can create momentum and incentives that other industry players will follow,” he told Reuters.
Until the identification system is perfect, Meta will ask Instagram and Facebook users to label doctored audio and video content. Violators of this policy will be subject to penalties.
One content created by AI that cannot yet be identified with technology is AI-generated text, such as that produced by ChatGPT. “That's in the past [that ship has sailed],” Clegg said.
Meta is also silent about its plans for AI-generated content spread across WhatsApp.
|
Photo: Arie Pratama
[THUMBNAIL] FB, IG, WA Down |
Deepfake scam
Deepfake content is now also starting to be used for online fraud. A finance worker at a multinational company was tricked into paying US$25 million (Rp. 392.97 billion) to fraudsters using deepfake technology.
According to Hong Kong police, the fraudster disguised himself using a deepfake as the company's chief financial officer in a video conference call.
The victim was deceived by being told to attend a video call which was said to be attended by several other staff members. But everything is actually a fake recreation, said Hong Kong police, quoted from CNN International, Monday (5/2/2024).
“(In) a video conference that was attended by many people, it turned out that everyone… [dia lihat] is fake,” senior watchdog Baron Chan Shun-ching told the city's public broadcaster RTHK.
Chan said the worker became suspicious after he received a message purportedly from the company's UK-based chief financial officer. Initially, the worker suspected it was a phishing email, because it contained a request to carry out a confidential transaction.
However, the worker put aside his initial doubts after the video call. Because, Chan said, the other people present looked and sounded like colleagues he knew.
Therefore, the worker agreed to send a total of US$200 million Hong Kong dollars or around Rp. 392.97 billion.
This case is one of many cases involving deepfake technology. At a press conference last Friday, Hong Kong police said they had made six arrests in connection with the scam.
[Gambas:Video CNBC]
Next Article
Facebook-Instagram Nests Online Gambling, Kominfo Reveals Data
(wur)