Cisco Integrates Anthropic Claude Mythos for Adaptive Red Teaming
Cisco integrated Anthropic's Claude Mythos model into its AI Defense product to help enterprises simulate sophisticated attacks and secure AI agent deployments.
At the Cisco Live APJC event in Melbourne, Cisco announced the integration of Anthropic's Claude Mythos model into its AI Defense security product. This update enables an adaptive red teaming engine that allows enterprises to simulate complex, multi-step attacks using frontier-level reasoning to find security gaps in AI applications and agents.
To maintain safety, Cisco runs the model within a controlled sandbox environment. The company also expanded AI Defense to support Amazon Bedrock and Amazon CloudWatch, providing customers with better visibility into token expenditure and return on investment.
DJ Sampath, senior vice president and general manager of Cisco’s AI Software and Platform group, stated that the integration is a "game changer" for partners securing AI deployments. The adaptive red teaming capabilities will first be available to a select group of enterprises before expanding to other qualified customers.
