"GenAI Models Are Easily Compromised"

Lakera reports that 95 percent of cybersecurity experts have low confidence in Generative Artificial Intelligence (GenAI) security. In addition, red team data suggests that anyone can easily hack GenAI models. Anyone can use GenAI-specific prompt attacks to manipulate the models, gain unauthorized access, steal confidential data, and more. This article continues to discuss key findings from Lakera's "2024 GenAI Security Readiness Report."

Help Net Security reports "GenAI Models Are Easily Compromised"

Submitted by grigby1