Reprinted from Wen Wei Po
Recently, the "Deepfake" (deepfake) incident in Korea was exposed, and the "Room N" case that is believed to be a new criminal method has reappeared. A number of topic words about "Korea AI deepfake" have aroused heated discussions among netizens.
According to Korean media reports, deepfake crimes against women are becoming rampant in Korea. Deepfakes are the use of AI face swapping to create pornographic images. Korea netizens found a flood of chat groups on the messaging app Telegram in which members created and shared sexually explicit "deepfake" images of underage girls.
In the process, the perpetrator did not hesitate to reveal the identity of the victim woman, threaten the victim, and demand money.
The former Democratic Party of Korea (DPK) that exposed the "Room N" incident
Park Ji-hyun, co-chair of the Emergency Countermeasures Committee, said,
The number of people currently involved in deepfake crimes nationwide
may reach 220,000 (including possible duplicates),
As a result, many women are trapped in anxiety and fear.
Because of the targeting of young women
Deepfake pornography is rampant,
The President of Korea urged the authorities to do more
to "eradicate" the epidemic of digital sex crime in the country.
How to quickly identify AI face-swapping scams?
Deep counterfeiting in 5 seconds makes AI face-swapping fraud nowhere to hide;
How to build a secure communication environment for the "old and young"?
AI technology is a powerful countermeasure to telecom fraud.
On September 8, the 2024 National Cyber Security Publicity Week officially kicked off in Nansha, and topics such as AI face swapping and voice swapping on personal privacy have become the focus of everyone's attention. A "portrait photo" or a close-up video of a person may become a new means and tool for criminals to commit crimes.
In the picture, the technician is demonstrating the "AI face swap" technology, which can make the person in the photo move with just a static photo of the target person.
Zhang Bo, a member of the Artificial Intelligence Security Governance Professional Committee of the China Cyberspace Security Association, said, "Although AI face swapping and voice swapping have certain technical thresholds and operating costs, if it is obtained by criminals, it may be used to impersonate others and carry out related criminal activities." ”
Falsifying indecent photos through "face swapping".
Shenzhen police smashed a criminal gang
AI face-changing, voice changing and other technologies
Once used by criminals,
It will also become a tool for extortion,
There have been many such cases in various places.
Not long ago,
Shenzhen police smashed a criminal gang
Not long ago, Mr. Wu, a citizen of Shenzhen, suddenly received an MMS message from an unfamiliar mobile phone number, and when he clicked on it, he saw that it was an indecent photo of him and a strange woman in a hotel room.
Mr. Wu: At that time, I received an MMS message with a nude photo of me. He threatened me to contact him and transfer money before what time, and I called the police as soon as I thought something was wrong.
After receiving the report, the Shenzhen police immediately investigated the clues provided by Mr. Wu. By combing through the report records, the police found that Mr. Wu was not the only one in the city who had the same experience, and the suspects' extortion methods were the same.
As the investigation deepened, the police found that this was a criminal gang headed by Wang, with a clear division of labor within the gang, some people were responsible for retouching pictures, some people were responsible for finding targets for crimes, and some people were responsible for sending text messages and making phone calls for extortion. After grasping a large number of clues, the Shenzhen police launched a unified network collection operation and successfully arrested more than 10 criminal suspects led by Wang.
In the suspect's computer, the police found a large number of indecent photos after compositing and changing faces.
Zhang Wei, police officer of the Pingshan Branch of the Shenzhen Municipal Public Security Bureau: The relevant evidence extracted from the suspect's computer, these pictures are the victim's picture information collected by the suspect from various channels, and then synthesized and processed to blackmail the victim.
At present, Wang and others have been prosecuted by the procuratorate. The case is under further proceedings.
What is the principle of AI face swapping?
How to identify the flaws in face swapping at a glance?
It is understood that at present, various acts of infringing on personal information against the "AI face swap" technology occur from time to time, and are abused in the gray field and even in the field of illegal crimes.
What is the principle of AI face swapping? How to identify the flaws of face swapping at a glance and prevent them?
Zhang Bo showed the reporter in detail the process and principle of AI face swapping, "You only need to input a positive static picture, and with the help of artificial intelligence technology, you can 'move flowers and trees' in real time." ”
He further introduced the principle of three major steps of AI face swapping: facial recognition tracking, facial feature extraction, and facial change fusion. "The overall principle is to accurately identify face images through computer deep learning, and then extract important facial features such as eyes, nose, mouth, etc., and then match, replace and fuse these features with the target face images."
Nowadays, the "AI face swap" technology has been relatively mature, and it can even realize real-time face swapping on video calls, although it is difficult to distinguish with the naked eye alone, but Zhang Bo pointed out that as long as you pay a little attention, "AI face swap" still has obvious flaws, and the "horse feet" are in the facial contours and light.
What are the key points to focus on in identifying "AI face swapping"?
"If you receive a video message, a video call, you should carefully observe whether the other party's facial contours are abnormally blurred and whether the light is normal," Zhang Bo said, "If necessary, you can ask the other party to quickly and sharply raise their heads, nod their heads, turn their heads, and then carefully observe whether the facial contours are abnormally blurred during this process."
In real life, how can we distinguish AI face and voice change?
Wang Xiaoqing, deputy manager of China Mobile's information security and management operation center, said that on the one hand, facial features are checked, expressions and movements are observed, as well as the matching degree of voice and mouth shape. On the other hand, it is necessary to observe the mood and context, verify the source of information, and ask the other person to cover their face with their hands or perform other facial movements to determine whether the picture is coherent.
However, Zhang Bo pointed out that
To prevent online fraud, we still need to start from the upstream.
First of all, it is necessary to prevent excessive leakage of personal information.
"In scenarios where facial information needs to be collected, citizens still need to pay extra attention to the network environment and the surrounding environment." In the case of a suspected AI face swap, you can choose to use a more direct method, such as making a phone call to the person concerned, to confirm whether the other party's identity is real.
Cybersecurity experts suggest that the most critical reason for illegal behaviors such as AI face swapping is the leakage of personal privacy information. Therefore, in the process of daily surfing the Internet, we should try to avoid excessive disclosure and sharing of personal biological information such as faces and fingerprints. If you receive a video of "family", "leader" and other transfers and remittances, you must carefully screen the authenticity.
Wary!
Don't post these kinds of photos casually
In daily life,
Many people like to be on the circle of friends and short videos
Bask in the food, bask in the beautiful scenery,
But if you're not careful,
These behaviors are likely
Indirectly divulging your personal information.
If it is obtained by someone with ulterior motives,
It may endanger yourself and your family
property and even personal safety.
Which photos must be careful,
Don't just post it on the Internet?
Do not post a photo with an ID card or a blank sheet of paper. These photos will contain sensitive information such as personal names, ID numbers, and addresses, and once leaked, criminals may use this information to commit identity theft, handle financial services such as credit cards and loans, and even carry out illegal activities. In addition, photos such as holding blank paper or newspapers are often used to verify identity, and if obtained by criminals, they may be used to make fake documents or carry out online fraud, which will bring financial losses and reputational damage to individuals.
Do not post photos of various ticket types. Many people on business trips and travels like to take train tickets or plane boarding passes and share them with their circle of friends. These tickets will not only contain information such as name, place of departure, destination, etc., but also contain passengers' personal information in the stripe code or QR code, which may be cracked and stolen by criminals.
Don't post photos of house keys, license plates, etc. Through big data and other technical means, the offender will analyze the specific location of the publisher at a specific time through these photos, and will also obtain the relevant living habits of the publisher. So, the best thing to do is not to show obvious items in the photo, such as house keys, license plate numbers, etc.
Protecting the privacy of individuals is paramount.
The public should be vigilant,
Avoid posting on the web
Photos that contain sensitive personal information.
Forward to remind friends and relatives to pay attention~
Source丨Guangzhou Daily, CCTV News, Cover News, Observer Network, Orange Persimmon News