AI Can Steal Passwords By Analyzing HDMI Cable Connections

August 14, 2024

The Three C’s: Collaboration Between Cybersecurity and Compliance

AI Can Steal Passwords By Analyzing HDMI Cable Connections

Cyber Security News reports that security researchers in Uruguay have discovered that hacker-deployed AI can steal passwords by decoding electromagnetic emissions from high-definition multimedia interface (HDMI) cable connections.

The good news for average home users of computers is that they are unlikely targets. Executing such an attack requires significant technical expertise and very specialized equipment, so the potential reward must be high.

The bad news for corporations and government agencies? They are not only good targets, but according to lead researcher Federico Larroca, “it’s likely that similar techniques are already being used by advanced hackers or state-level actors.”

Larroca says an attacker could potentially capture HDMI emissions from outside a building using equipment hidden in a nearby vehicle.

The level of accuracy attained by the technique is high enough for an attacker to steal sensitive data including any encrypted communications displayed on a screen, including passwords. Larroca’s research team’s AI system improved on previous eavesdropping attempts by more than 60 percentage points.

Researchers have long been aware that similar vulnerabilities existed for older cathode ray tube monitors, but the complex digital signaling used by modern HDMI connections was thought to make such attacks impractical. The new research demonstrates that AI can overcome complex signaling.

Cybersecurity experts recommend avoiding proximity to windows, working in public spaces, and being aware of your surroundings when dealing with sensitive information. Using encrypted connections and not displaying passwords on external monitors connected via HDMI may also be helpful in mitigating risks.

Read full article at:

Get our free daily newsletter

Subscribe for the latest news and business legal developments.

Scroll to Top