Feature Anthropic has positioned itself as a leader in AI security, and in a recent analysis by Chatterbox Labs, it proved to be the case.
AI vs Machine Learning
Chatterbox Labs tested eight major large language models (LLMs) and all were found to produce malicious content, although Anthropic's Claude 3.5 Sonnet fared better than the competition.
The UK biz offers a suite of tests called AIMI which rates LLMs on various 'pillars' such as 'fairness', 'toxicity', 'integrity' and 'security'.
"Security" in this context refers to model security—resistance to broadcasting malicious content—rather than the presence of potentially exploitable code flaws.
