Cisco Launches AI Security Leaderboard, Anthropic Dominates as Mistral and DeepSeek Fall to the Bottom
Summary
Cisco launches its LLM Security Leaderboard, revealing Anthropic's Claude models dominate AI safety rankings with 8 of the top 10 spots, while Mistral, DeepSeek, and other major providers fall dangerously to the bottom in resistance to adversarial attacks.
Key Points
- Cisco has launched its LLM Security Leaderboard, a publicly accessible ranking that evaluates AI models based on their resistance to single- and multi-turn adversarial attacks, aiming to help enterprises assess AI security risks.
- Anthropic dominates the rankings by claiming 8 of the top 10 spots, with Claude Opus 4.5 taking first place, while OpenAI's GPT-5.2 and GPT 5 Nano land in 7th and 9th place respectively.
- Mistral's models rank at the very bottom of the leaderboard, with other low-scoring models coming from DeepSeek, Cohere, Qwen, and xAI, highlighting significant security gaps across several major AI providers.