Cloudflare has announced two new security solutions designed to help protect against AI-based threats: Defensive AI, a personalized approach to securing organizations against risks caused by AI, and Firewall for AI, which scans LLM prompts for language commonly used in attacks. 

Defensive AI provides tailored mitigations that enable the protection of critical applications and entire networks.

“Fighting AI with AI is now a non-negotiable. And a personalized approach to protect data and defend against complex threats unique to an organization’s attack surface, at speed, and scale, is paramount,” said Matthew Prince, CEO and co-founder at Cloudflare. “By understanding ‘normal baselines’ in a customer’s environment and mitigating the threats that will move the needle towards increased resilience, Defensive AI is the crucial edge defenders need to stay ahead of today’s adversaries.”

Cloudflare’s Defensive AI offers a customized defense strategy by analyzing specific customer traffic patterns, leading to enhanced protection tailored to each organization’s unique digital environment. 

This feature aims to safeguard applications by learning their typical behavior and identifying deviations that could indicate attacks, thus protecting against damages, account takeovers, or data theft. Additionally, Cloudflare is addressing the primary threat vector in cybersecurity: email. 

By utilizing AI to scrutinize email content for suspicious elements, Cloudflare’s Cloud Email Security solution is designed to counteract the increasingly sophisticated phishing attacks made possible through advanced AI, ensuring a more resilient defense against cyber threats.

Cloudflare is also addressing the challenge of insider threats, which often involve privileged employees potentially compromising security intentionally or accidentally. By implementing a Zero Trust approach via Cloudflare Gateway, organizations can monitor and evaluate user behavior and resource access. This system flags risky or unauthorized activities by establishing a behavior baseline, allowing for a nuanced assessment of both user and resource trustworthiness. 

Through these initiatives, Cloudflare’s Defensive AI is set to significantly bolster organizational cyber defenses by addressing critical vulnerabilities in web applications, email security, and insider threat mitigation, aligning with contemporary cybersecurity needs.

Cloudflare’s second new security solution, Firewall for AI, is designed to enhance protection for AI applications by scanning API requests containing LLMs. 

This approach mirrors the deployment of traditional Web Application Firewalls (WAF), where incoming requests are meticulously analyzed for any patterns or signatures that could indicate potential security threats. This innovative firewall aims to safeguard applications against a range of attacks, ensuring the integrity and security of AI-driven processes.

The Firewall for AI is versatile in its deployment options, offering protection for models hosted directly on Cloudflare’s Workers AI platform as well as those running on third-party infrastructures. It integrates seamlessly with the Cloudflare AI Gateway, providing a robust security layer, according to the company in a post.