Nvidia today rolled out artificial intelligence (AI)-powered digital fingerprinting capabilities as part of its AI Enterprise 3.0 platform, aiming to deter identity attacks.

Digital fingerprinting technology is designed to detect anomalies by learning the typical network behavior of organizations, users, systems, and applications and then mapping how they interact with complex enterprise infrastructures day-to-day, the vendor claims.

“With Nvidia digital fingerprinting technology, cybersecurity teams can get instant alerts on high-risk behavioral changes for users that are on their network,” Nvidia VP Justin Boitano said during the pre-briefing. “By using AI in cybersecurity, AI can watch all the login and data access events of a user across the organization and alert security teams on behavioral changes that would indicate that the user's credentials may have been compromised.”

Boitano claims the Nvidia Digital Fingerprinting AI Workflow can help security teams reduce the amount of data and security events that they need to manually sift through to only a handful of high-risk alerts that they can deeply analyze and then mitigate the threats.

“Our approach is we build highly tuned models for the behavior of every user, so that if somebody steals my credentials, and suddenly that user starts going after build systems or starts going after IT systems trying to create users with super admin privileges. It's an alert that's relevant to our security teams,” he explained. “so we alert on anti-patterns of behavior to the security team.”

Looking ahead, the Nvidia R&D team will work on using AI and accelerated computing for securing digital identities and generating hard-to-find training data.

Bartley Richardson, director of cybersecurity engineering and R&D at Nvidia, forecasts a detailed digital identity model — that knows how fast a person types, with how many typos, what services they use, and when they use them — will replace passwords and multi-factor authentication and prevent attackers from hijacking accounts and pretending they are legitimate users.