Terra Security Launches AI Native Cybersecurity Podcast With Inaugural Guest Amazon CISO CJ Moses

Terra Security has launched the AI Native Cybersecurity Podcast, a new series co-hosted by co-founder and CEO Shahar Peled that brings together security leaders who are putting artificial intelligence to work on both sides of the fight. The inaugural episode features CJ Moses, CISO of Amazon, in a wide-ranging conversation about how generative AI has reshaped the threat landscape and what defenders must change in response.

The show describes itself as field notes from security teams defending against and operating like AI adversaries, built on the conviction that the best defense is a good offense. Peled positioned the series as a home for thought leadership from the most innovative cybersecurity leaders working with AI, and he opened with one of the most prominent security executives in the world. The podcast debuted with a preview on September 30, 2026, followed by its first full episode featuring Moses on October 18, 2026.

A Guest With Deep Roots at Amazon

Moses leads security engineering and operations across Amazon, with a mission of enabling the company’s businesses by making the benefits of security the path of least resistance. He joined Amazon in December 2007 and held several roles, including Consumer CISO and AWS CISO, before becoming CISO of Amazon in September 2023.

His remit covers what Amazon calls integrated security. Cyber, personnel, and human intelligence are not treated as separate disciplines. As Moses put it, “We pull it all together into one.” That single function spans more than 148 business units, so information held anywhere in the company can be connected in time to stop harm before it happens.

Before Amazon, Moses led technical analysis of computer and network intrusion efforts at the FBI’s Cyber Division and served as a Special Agent with the Air Force Office of Special Investigations. He holds degrees in Computer Science and Criminal Justice and competes as an SRO GT America GT2 race car driver.

Speed Defines the New Landscape

Asked to name the single biggest change since generative AI went mainstream, Moses pointed to “The speed at which you need to respond to them.” The familiar rhythm of 30-, 60-, and 90-day patching windows was still the prevailing mindset as recently as six months ago. With AI systems such as Mythos demonstrating new vulnerability-discovery capabilities, Moses argued that exposure windows are collapsing toward seconds.

Amazon sees this firsthand through MadPot, its large honeypot infrastructure. In one observed case, an adversary moved from initial intrusion through reconnaissance, lateral movement, and action across every reachable system in 12 minutes and 42 seconds. A comparable campaign once played out over weeks or months.

From Hiring Pipelines to Shared Responsibility

The episode also explored how AI is changing insider threats. Moses described North Korean operators using AI to generate personas and backstop entire identities at industrial scale in pursuit of jobs at companies like Amazon. Amazon has tracked the activity as far back as April 2024, and Moses noted that “they’re playing a long game.” To date, Amazon has stopped at least 2,900 DPRK actors during the hiring process.

Moses also revisited the AWS shared responsibility model, a framework he helped create in 2008. He explained that it continues to hold in the AI era, including for services such as Bedrock, because it was written from the start to define where the provider’s responsibility ends and the customer’s begins.

Defense That Keeps Pace

Moses outlined how Amazon is matching adversary velocity. Code is scanned iteratively as it is written within faster CI/CD pipelines, followed by a final scan and penetration testing of the finished product. Holding back development to add caution, he explained, works against defenders: “The reality is if you slow down, you’re actually playing into the hands of the thing we talked about in the beginning.” AI-assisted patching has expanded coverage many times over in recent months, and MadPot telemetry helps the team decide what to fix first.

The conversation covered continuous offensive testing as well, an area at the heart of Terra’s work. Because AI can chain vulnerabilities together, the traditional math of stacking defensive controls no longer holds on its own. Amazon continually tests its own environment and acts on what it finds before anyone else can.

The Metric That Will Matter

For his closing prediction, Moses offered a view he expects to define security programs over the next 12 months: “The measure of a good security program won’t be how many vulnerabilities you found.” He sees mean time to defense, the window between discovering a vulnerability and patching or otherwise mitigating it, becoming the measure that matters most. A small group at Amazon has already made it their standard.

More Episodes Ahead

Peled invited listeners to send in their questions, and Moses said he would “look forward to coming back again when things are inevitably going to change. We’re gonna have more fun things to talk about.” The first episode sets a clear direction for the series: practical, operator-level insight from leaders already using AI to defend against AI.