Leading AI models are more vulnerable to malicious prompts than vendors claim

Cisco’s evaluation of 15 leading AI models from OpenAI, Anthropic, Google, Amazon and xAI found that single-turn attack success rate (ASR) is not a reliable proxy for what happens when an attacker can adapt across turns. Tests conducted by researchers Nicholas Conley and Amy Chang revealed that AI models were much more susceptible to multi-turn [&hellip The post Leading AI models are more vulnerable to malicious prompts than vendors claim appeared first on TechCentral.ie.