# Amazon CISO CJ Moses on AI-Driven Cyber Threats Amazon's security lead discusses how AI accelerates attack velocity and changes defense strategies. - **Source:** [The Next Web](https://brifing.si/source/afe24b20-b275-4c19-b9a6-8bb3c8babae7) - **Published:** 2026-10-20 - **Topic:** [Software & Engineering](https://brifing.si/c/engineering) - **Original article:** - **On Brifing:** https://brifing.si/s/fa1d6b07-8d16-4721-adf9-7a4e175e517d ## Accelerated Attack Timelines Amazon CISO CJ Moses highlights a critical shift in threat landscapes during the AI Native Cybersecurity podcast. Attackers now transition from initial access to full system takeover in mere minutes. Specifically, observed incidents show attackers compromising systems in just 12 minutes and 42 seconds, a process that previously required weeks or months. ## Velocity as Primary Metric The core challenge for modern security teams is the increased velocity of adversaries enabled by artificial intelligence. Traditional defensive strategies relying on long patching windows are becoming obsolete. Security leaders must now prioritize response speed over the sheer volume of vulnerabilities discovered, as attackers exploit weaknesses faster than conventional systems can react. ## Redefining Defensive Controls While layered defensive controls remain relevant, they are insufficient against AI-enabled adversaries who chain individual weaknesses into complex attack paths. Organizations must shift toward continuous testing environments. This approach helps security teams determine which vulnerabilities are truly exploitable, allowing for more effective resource allocation rather than chasing every single identified system weakness. ## Integrating Security into Pipelines Amazon maintains development speed by embedding security scans directly into continuous integration and delivery pipelines. AI-assisted scanning identifies issues during code development, while subsequent testing validates the final product. This integration ensures that security measures do not unnecessarily delay software delivery, effectively balancing the need for speed with robust system protection. ## Intelligence-Led Patching Priorities Amazon utilizes its MadPot honeypot infrastructure to gather real-time intelligence on adversary behavior. This data informs patching priorities by highlighting which vulnerabilities attackers are actively targeting. By understanding the specific methods and targets of adversaries, security teams can focus their efforts on mitigating the most critical risks before they are exploited. ## Identity-Based AI Threats AI risks extend beyond technical infrastructure vulnerabilities to include sophisticated identity-based deception. Amazon has identified and blocked at least 2,900 North Korean actors attempting to infiltrate the company through fabricated identities and AI-generated personas. This highlights the growing intersection between cybersecurity, personnel security, and hiring processes in an AI-driven world. ## Cross-Functional Security Models Amazon employs an integrated security model that connects cyber intelligence, personnel security, and internal investigations across business units. This structure prevents warning signs from remaining isolated within specific organizational functions. As AI scales deceptive operations, the ability to synthesize information across different departments becomes essential for detecting and neutralizing complex, multi-layered threats. ## Shifting Performance Metrics The industry must move away from measuring success by the number of vulnerabilities patched. Instead, focus should shift toward the time elapsed between discovery and mitigation. Whether through blocking access or system isolation, the goal is to close the window of opportunity for attackers, effectively reducing exposure to real-world threats. ## Sources covering this story - https://thenextweb.com/news/terra-security-ai-native-cybersecurity-podcast-cj-moses-amazon — --- _Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._