Computerra: Can AI Protect User Accounts?

2025-01-15 14:55

Cybersecurity experts agree: passwords remain one of the weakest links in data protection. According to Kaspersky Lab, more than half (59%) of tested passwords can be cracked in under an hour with nothing more than a modern graphics card and basic technical knowledge. Hackers continually find new ways to steal data, while users often choose overly simple passwords to avoid memorizing complex combinations. The future, however, may lie in artificial intelligence.

Alexander Vorfolomeev, Head of IT at Selecty, explains:

«AI technologies are already being actively applied in multi-layer authentication. For example, machine learning–based systems can analyze behavioral factors such as typing patterns, mouse movements, or how a user interacts with an application. Biometrics is also becoming more accurate thanks to AI — from fingerprint and retina scans to facial recognition. Still, by 2025, AI is unlikely to completely replace passwords. They will remain a familiar fallback option for users».

In the near future, algorithms will be able to detect suspicious activity in real time, greatly reducing the risk of account breaches. At the same time, protecting the AI systems themselves will become essential, as they too will become prime targets for cybercriminals. Businesses are expected to increase investment in employee training to ensure proper use of these tools and to rethink their data governance policies with greater emphasis on privacy.

Vorfolomeev sees promise in FIDO2 (Fast Identity Online) technology:

«These solutions eliminate passwords entirely by using public-key cryptography, and they have the potential to become the standard in the coming years. The key factor for their success will be achieving the right balance between convenience and security».