KAIST Develops AI That Finds Its Own Hidden Weaknesses, Paving the Way for Safer Generative AI Models (IMAGE)

The Korea Advanced Institute of Science and Technology (KAIST)

KAIST Develops AI That Finds Its Own Hidden Weaknesses, Paving the Way for Safer Generative AI Models

Caption

Figure 3. Graph comparing the Attack Success Rate (ASR) and attack diversity (# of unique attacks) between S-GFN and existing methods. Stable-GFN increases attack diversity by a factor of seven compared to standard GFNs, all while maintaining an ASR of over 90%.

Credit

KAIST

Usage Restrictions

No

License

Licensed content

Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.