The AI expert named ways to protect personal data from leakage

Artificial intelligence (AI) cannot hack into people's accounts, it correlates human behavior, metadata, and duplicate markers, and by knowing the AI's behavior, data leakage can be avoided. This was announced on September 26 by Igor Nikitin, an expert in the field of AI, the founder of WMT Group and WMT AI.

"For example, you can look at a photograph: reflections in glass, badges, QR codes. Find out the metadata of a photo: where, when, and on which device it was taken. In addition, neural networks analyze human behavior: when they most often post, which routes they take. And his timbre and manner of speech are biometric data that deepfakes successfully work with," he said. "Газете.Ru ".

According to the expert, storing documents in paper form will help prevent information leakage to the Network. Medical, financial documents, contracts, copies of passports and other identification documents — none of this should be in the cloud.

If it is necessary to publish any information, you need to make sure that there is no personal data in it. Nikitin recommends deleting geometries, EXIF data (hidden information about the camera, date and location of shooting), as well as blurring the background and other people's faces and not recording audio messages that are too long to avoid deepfakes.

Nikitin advised using different passwords, email addresses, and phone numbers for different areas of life. In his opinion, this will help confuse the AI. For example, one phone is for family and friends, the other is for work calls, and so on.

You can also put a honeytoken on digital documents, a data element that is intentionally created as a bait or trap for cybercriminals.

He summarized that the camera, microphone, and geolocation should be set to "in-use only" mode, and notification previews on the locked screen should be minimized. The laptop requires additional security measures. So, the expert recommended using a physical shutter for the webcam, enabling automatic system updates and activating disk encryption.

Yandex.Weather in the domain: fraudsters have begun to actively substitute payment details for companies

The number of such incidents has increased by a quarter in a year.

On September 19, cybercrime specialists reported that fraudsters began stealing money from participants in a special military operation (SVO) and their families by creating duplicate websites for charitable foundations and offering participation in a fictitious investment program with monthly "support" of up to 500 thousand rubles. According to experts, scammers demand to make a contribution of 50 thousand rubles and promise income of up to 21% per month.

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