An IT expert gave tips on recognizing deepfakes

Deepfakes have evolved from entertainment into a serious tool for committing crimes. Alexey Markelov, director of the AI Competence Center at ANO Digital Platforms, told about this on Tuesday, February 3.

"It's becoming more difficult to distinguish a deepfake video generated by scammers from a live person, but for now it's still possible thanks to the technical artifacts that the neural network leaves on the material," he said in an interview with RT.

According to him, qualitatively new models appeared last year, creating fully synthesized content without the original video, which requires an integrated approach — legal, organizational and technical.

Markelov said that visually problems in deepfakes usually manifest themselves in facial expressions and dynamics: the face looks too "smoothed out", emotions are simplified, and blinking is unnatural.

He clarified that deepfake may have distortions around the edges of the face in the area of the eyes, mouth, hair or chin when turning his head or making sudden movements.

"The slow-down of the video is especially significant: at low speed, there are noticeable interruptions in the synchronization of speech and lip movements, which are almost not found in live video. Even if the video looks perfect, the voice may be robotic, with unnatural pauses, with a violation of the rhythm of speech," he explained.

Markelov added that it is necessary to ask the interlocutor to pronounce a phrase with a complex intonation in order to recognize the fraudster.

Naked in the wires: how to profit from strawberry lovers on the Web

Scammers strike at Fetishists

In January, experts in Russia and abroad noted the growth of deepfakes created on the basis of photos and videos taken in 2016, which users massively posted on social networks. Cybersecurity experts told Izvestia that such archival materials are publicly available and are quickly collected by fraudsters using automated programs to create plausible fakes.

Earlier in the day, the Paris prosecutor's office invited American entrepreneur and billionaire Elon Musk to testify after searches of the offices of his social network X (ex. Twitter), according to the website kp.ru .

The rest of the company's employees were also called to the "free hearings" as witnesses from April 20 to 24, IA Regnum reports.

On the same day, it became known that the French prosecutor's office began conducting searches in the offices of the social network X. It is noted that the searches took place jointly with employees of the National Cyber Division and Interpol.

Earlier, on January 26, the European Union (EU) launched an official investigation against Mask's company xAI after the scandal surrounding the Grok chatbot, which "undresses" people based on photos, reports 360.ru . If it turns out that the company is violating the rules, the EU can impose a fine of 6% of the global annual turnover.

All important news is on the Izvestia channel in the MAX messenger.