AI benefits/risks

There are a lot of juicy and theoretical AI cybersecurity risks. Let's look at the real ones and how to mitigate them for a change.

Mounting artificial intelligence-related cybersecurity risks have prompted cybersecurity agencies from the Five Eyes countries, which includes the U.S., to release new joint guidelines regarding the safe deployment and operation of AI systems, according to SecurityWeek.

The “Crescendo” attack uses a chain of seemingly benign prompts to achieve an adverse output.

A PowerShell script used to deploy the infostealer contains unusually specific comments, researchers say.

While the employee did not fall for the scam, LastPass took the incident as an opportunity to spread awareness about deepfakes.

No, it's not safe, but now’s the time to take proactive steps and adapt.

Here are four insights into how security pros can judge new AI products when vendors say they were “battle-tested in Ukraine.”

New joint research by Wiz and AI-as-a-service provider Hugging Face find that a malicious pickle-serialized model could contain a remote execution payload.