An AI specialist warned of the dangers associated with neural networks

By uploading documents to public neural networks, the user loses control over them. Yuri Silaev, head of the Laboratory of Trusted Artificial Intelligence at RTU MIREA, warned about this on Tuesday, July 14.

First of all, this concerns personal data, financial papers and medical reports, he stressed.

"When working with neural networks, it is worth adhering to the principle of minimum sufficiency: transfer to the service only the data that is really needed to complete the task," the specialist explained in a conversation with Lenta.Ru .

You cannot upload documents with personal data to neural networks — photos of your passport, driver's license, SNILS, INN, bank cards, insurance policies, tickets with QR codes, as well as any information about your residential address, date of birth, document numbers and phone numbers.

Silaev added that financial and legal documents should also not be transferred to AI services without first clearing the data. Account statements, contracts, powers of attorney, and tax returns contain a large amount of confidential information, he said.

The expert added that medical documents — test results, medical history extracts, doctors' reports, CT and MRI scans — contain not only information about the patient's health, but also personal data. Therefore, if necessary, you should delete the last name, date of birth, policy number, barcodes and QR codes before uploading.

He stressed that internal work correspondence, commercial proposals, draft contracts, source code, customer bases and financial reports should also not be released into publicly available neural networks — this threatens both leakage and violation of contractual obligations.

"If it is impossible to do without uploading a document, you should delete the names, addresses, document numbers, bank details, QR codes, signatures and names of organizations before sending them. The less sensitive data is transmitted to the AI service, the lower the risk of its further dissemination or misuse," Silaev concluded.

Selective model: foreign neural networks will not be banned in Russia

How will this affect the implementation of artificial intelligence in its regular users?

In turn, the press service of the Moscow Region Ministry of State Administration warned that publishing personal information on social networks could attract the attention of intruders. 360.ru .

Sergey Taran, Managing Director of LANSOFT, told Izvestia on July 13 that digital sustainability is becoming one of the key factors of business competitiveness, going far beyond the tasks of IT departments. Companies that are able to quickly recover from failures and continue fulfilling their obligations gain a significant advantage over their competitors.