Cisco's AI Defense: Tackling the Growing Threat of Adversarial AI

The Growing Threat of Adversarial AI

As artificial intelligence (AI) adoption accelerates across enterprises, a new security paradox emerges: how to protect a system that constantly evolves while scaling it enterprise-wide. This challenge is at the heart of a stealth cyber war, where adversarial AI is dominating the threat landscape. Adversaries are quick to weaponize every aspect of AI, including large language models (LLMs), opening up new attack surfaces that traditional security technologies struggle to defend against 1

1

.

The gap between adversarial AI and defensive AI is growing rapidly, putting enterprises' security and financial stability at risk. From data poisoning to prompt injection attacks, adversaries are already exploiting AI's vulnerabilities, turning the technology into a vector for misinformation, security breaches, and business disruption.

Cisco's AI Defense Strategy

To address these challenges, Cisco has launched its AI Defense strategy. With Gartner predicting that 40% of generative AI implementations by 2028 will lack adequate security coverage, Cisco's timing is crucial 1

1

. The AI Defense platform, launched in January, integrates real-time monitoring, model validation, and policy enforcement at scale.

Jeetu Patel, EVP and CPO at Cisco, emphasizes the importance of this approach: "Business and technology leaders can't afford to sacrifice safety for speed when embracing AI. In a dynamic landscape where competition is fierce, speed decides the winners." 1

1

Key Features of Cisco AI Defense

Cisco's AI Defense is purpose-built, embedding security into network infrastructure to protect every aspect of AI development, launch, and use. The platform offers:

1. Real-time monitoring of AI models and applications
2. Automated model validation to ensure AI behaves as intended
3. Policy enforcement at scale to prevent unauthorized AI use
4. Integration with existing network infrastructure for seamless deployment

Addressing AI Supply Chain Security

In addition to its AI Defense platform, Cisco is tackling the critical issue of AI supply chain security. The company has announced that all existing users of Cisco Secure Endpoint and Email Threat Protection are now protected against malicious AI Supply Chain artifacts 2

2

.

This protection covers artifacts downloaded from open-source repositories like Hugging Face, shared via email, or downloaded from shared drives. The solution offers:

1. Automatic detection and blocking of malicious AI artifacts
2. Protection against model hijacking, backdoors, and other AI-specific threats
3. Integration with existing Cisco security products for comprehensive coverage

The Importance of AI Supply Chain Security

AI Supply Chain Security encompasses practices to protect enterprises throughout the AI development and deployment process. This includes securing software stacks, training data, and third-party models against vulnerabilities such as software flaws, deserialization issues, and data/model poisoning 2

2

.

Organizations face several challenges in securing their AI supply chain, including:

1. Balancing innovation with security concerns
2. Managing the complexity of AI model dependencies
3. Ensuring the integrity and security of training data
4. Protecting against model theft and intellectual property loss

Future of AI Security

As AI continues to evolve and integrate into enterprise systems, securing the AI supply chain becomes increasingly critical. Cisco's approach demonstrates that organizations need not sacrifice security for innovation. With ongoing research and development in AI security, including contributions to open-source projects like the AI Risk Database, Cisco is positioning itself as a leader in the growing field of AI security 2

2

.