- Статьи
- Internet and technology
- Compromise in three moves: how neural networks are changing the schemes of online blackmailers
Compromise in three moves: how neural networks are changing the schemes of online blackmailers
Artificial intelligence (AI) is changing the schemes of online blackmailers directed against Russians, experts have warned about this. Thanks to neural networks, attackers no longer need to extract real compromising material: now they can easily produce fake sensitive materials based on ordinary images from open sources. For more information about how AI is changing the tactics of online scammers, how dangerous the updated schemes are and how to protect yourself from them, read the Izvestia article.
Why are AI technologies interesting to online blackmailers
Modern AI technologies attract network blackmailers by radically changing their schemes — attackers no longer need to have real compromising evidence, Alexander Vurasco, Director of Development at the Solar AURA External Digital Threat Monitoring Center (Solar Group), says in an interview with Izvestia.
"Blackmailers no longer need to hack into a phone, gain access to a personal archive, or convince a person to send an intimate photo, since a fake can be made based on ordinary pictures from open sources," the expert explains. — Therefore, the argument "I've never filmed anything like this, so I have nothing to fear" here, unfortunately, does not work.
Today, the quality of deepfakes is growing, and the characteristic AI features are becoming less noticeable, adds the deputy head of the department of audits and compliance with information Security requirements at the Ural Center for Security Systems (UCSB) IT company Ksenia Kuznetsova. Thanks to this, blackmailers can easily scale attacks and quickly influence the victim, since the content is not extracted, but synthesized by the model.
At the same time, as Alexander Vurasko notes, attackers do not always need flawless fakes. Their task is to make a potential victim afraid that compromising files will be seen by relatives, colleagues or acquaintances.
What schemes with compromising information from neural networks to expect at the end of 2026
In the coming months, targeted attacks by online blackmailers may be further developed — more and more detailed real information about potential victims will be added to the generated compromising material, says Vladislav Shelepov, GSOC threat analyst at Gazinformservice.
"For example, a fraudster may know where the victim works and threaten to send a fake video to colleagues," the source points out. — Attackers can often find such information in open sources and in leaked databases.
The threat of "we'll send it to everyone" is abstract, and mentioning a specific relative or employer makes the pressure more convincing, agrees Alexander Vurasko. At the same time, according to the expert, the topic of such blackmail will not necessarily be intimate at all — fake voice messages or video recordings are possible in which a person allegedly confesses to an unseemly act or discusses illegal actions.
Then the potential victim will be offered to pay for the refusal to publish. Another risk is repeated deception under the guise of assistance, the expert says. Under such schemes, after threats, a person may be offered for money to "find a scammer" or "guaranteed to delete all copies." Therefore, the promise of saving from blackmail must also be verified.
— Another possible scenario is the use of fake voice messages or videos in conjunction with ordinary blackmail, — notes Vladislav Shelepov. — This can make the threats more convincing and create the impression that the attacker has additional materials. It is also possible that more widespread schemes may appear, when criminals automatically create compromising material based on photos from open profiles and send out demands for money.
At the end of the year, according to Ksenia Kuznetsova, debtors may turn out to be a separate category of potential victims of online blackmailers: they will be reminded of their debts, while exerting pressure with fake images. Targeted attacks on company employees with the generation of fake compromising materials to gain access to corporate systems or data are also possible.
What schemes with AI compromising material were previously encountered on the Web
Among the attacks with AI compromising materials that have been encountered before, Sergey Shcherbakov, technical director of the Stakhanovets company, an expert in the field of information security and IT, highlights cases of blackmail using synthesized images created on the basis of open photographs. Everyday images, including beach and home shots, served as the source material for generation.
"Episodes of systematic mailing of intimate images with the victim's face, accompanied by financial demands, were also recorded," the source tells Izvestia. — A case of death of a teenager as a result of such blackmail has been recorded abroad.
A 32-year-old woman in Moscow recently became a victim of the use of AI compromising materials: according to a REN TV source, she contacted the police after she was threatened with intimate photos generated by a neural network due to loan delays. The victim had overdue payments in August and was thinking about bankruptcy. Since October, unknown people began to write to her daily in messengers, after which the woman contacted the police.
Meanwhile, according to Alexander Vurasko, such pressure can be directed at a person of any gender and age. It is important for an attacker to find what a particular victim is afraid of: family reactions, problems at work, or harassment among acquaintances. You don't have to be a public figure to do this — a small circle of people whose opinions matter is enough.
— Completely fake images can cause problems at work or lead to harassment, — confirms Vladislav Shelepov. — Moreover, after the first transfer to the extortionist, blackmail does not necessarily stop. On the contrary, the attacker understands that the pressure is working, and may demand even more money.
How to protect yourself from blackmail attacks using neural networks
In order to protect themselves from online blackmailers using AI technologies, experts interviewed by Izvestia advise following certain security rules. However, according to Alexander Vurasko, it is impossible to completely eliminate the risks of creating a fake with someone's face. However, you can restrict access to personal photos, a list of friends, and other information that makes it easier to prepare an address pressure.
"Unique passwords and two—factor authentication will help protect personal correspondence and archives from account hijacking," the expert notes. — At the same time, it is important to understand the limits of such protection: it does not prevent the use of a photo that is already in the public domain.
If threats have already been received, you should not pay for silence, send new images or fulfill other requirements, warns the interlocutor of the editorial office. Payment does not guarantee the removal of materials and may lead to continued blackmail. In such cases, Kaspersky Lab's chief expert, Sergey Golovanov, recommends contacting law enforcement agencies and a lawyer.
At the same time, it is important to preserve the evidence: notarize the correspondence, record the images and other materials received, and make copies of them. In the future, the victim can defend his honor and dignity in court, including in connection with the dissemination of defamatory materials.
One of the important measures to counter such abuses could be the mandatory labeling of content created using AI, believes Sergey Golovanov.
— At the corporate level, regular trainings for employees on recognizing social engineering and deepfakes are required, as well as a clear incident response policy with an accessible communication channel with information security and lawyers, — concludes Ksenia Kuznetsova.
Переведено сервисом «Яндекс Переводчик»