The risks of cheap deepfakes: From darknet ads to calls for vigilance

DNVN - A group of cybersecurity experts recently discovered many deepfake video and audio creation services being advertised on the darknet at deeply discounted prices - up to 400 times lower than before.

With starting costs of only about 1.3 million VND, fake tools such as face transplant and voice imitation are raising warnings about the risk of fraud, fake personal images and cyber security consequences.

(Illustration)

Cheap prices on the darknet, hidden great danger

Kaspersky’s Global Research & Analysis Team (GReAT) has just published findings showing that many advertisements offering services to create deepfake videos and audio in real time have appeared on darknet markets. Starting prices are recorded at around 50 USD (about 1.3 million VND) and 30 USD (about 800,000 VND) for fake voice messages, depending on the complexity and length of the content.

Previously, “legitimate” deepfake services cost between $300 and $20,000 per minute.

The sharp drop in price – 400 times lower than the previous popular level – makes the risk of being exploited for bad acts even higher.

Advertising features and “bait” of bad guys

Ads now go beyond simple fake videos to include real-time face swapping during video calls, spoofing device cameras, or “replacing” faces to bypass authentication. They also advertise the ability to sync facial expressions with voiceovers, and even support multiple languages.

In addition, “voice cloning” services are also being sold, allowing voice imitation with the ability to adjust pitch and tone to express different emotions. However, many experts suspect that these are actually “customer-baiting” advertisements - buyers may be tricked into transferring money in advance, but not receiving the quality product as promised.

“The need to use this tool is also increasing”

“We have not only seen advertisements offering deepfake-as-a-service, but also noticed an increase in the demand for these tools,” said Dmitry Galov , Head of Kaspersky’s GReAT team (Russia and CIS) .

Cybercriminals are experimenting with and incorporating AI into their attack strategies, he said, with some platforms even developing their own “malicious” LLMs, separate from public models and able to run directly on the bad guys’ devices.

While deepfakes cannot yet be considered a completely new cyber threat, the technology could make attacks more sophisticated and harder to detect.

Recommendations and how to identify deepfake

Kaspersky recommends that businesses and organizations need to deploy comprehensive cybersecurity measures, not only stopping at installing protection software but also building a team of IT experts with the skills to detect and respond.

Additionally, staff within the organization should be trained to recognize unusual signs in video or audio that may be a product of deepfake technology. These videos often have jerky and unnatural movements, inconsistent lighting or skin tones between frames, characters appearing with unusual eye blinking frequency or almost no blinking, and faces that may be distorted or slightly deformed.

Additionally, image quality is often poor and frames are unstable, especially in low light, making the overall video look less realistic. Being vigilant about deepfake threats and raising awareness is seen as an important first step to minimizing damage from sophisticated scams in the AI ​​era.

Nguyen Bach

Source: https://doanhnghiepvn.vn/chuyen-doi-so/rui-ro-tu-deepfake-gia-re-tu-quang-cao-darknet-den-loi-keu-goi-canh-giac/20251016105304840