Artificial intelligence is changing the economics of cybersecurity, allowing attackers to identify weaknesses faster while giving defenders new ways to test and protect their environments. But the biggest shift may be happening in how organizations understand risk: moving beyond individual vulnerabilities toward the attack paths that connect them. Terra Security Co-Founder and CEO Shahar Peled discussed this transition with former Google CISO Gerhard Eschelbeck.
In a recent episode of the AI Native Cybersecurity podcast, Eschelbeck shared his perspective on AI-powered attacks, automated penetration testing, and the opportunities emerging for a new generation of cybersecurity companies.
AI Is Changing the Scale of Cyber Threats
With nearly 30 years in cybersecurity, including leadership roles at McAfee, Qualys, and Google, Eschelbeck has seen multiple technology cycles reshape the industry. He believes AI is introducing an unprecedented level of speed and scale, while also creating entirely new security challenges.
“The speed, the scale, the amount of vulnerabilities and risks that we are seeing is unheard of,” he said, pointing to emerging threats such as prompt injection alongside the growing volume of traditional vulnerabilities.
However, the most important change may be how attackers combine existing weaknesses. Rather than exploiting a single vulnerability, they can chain multiple issues together to create more sophisticated attack paths.
“What has changed is the chaining of vulnerabilities,” Eschelbeck explained. This makes it increasingly important for security teams to understand how vulnerabilities interact rather than relying exclusively on individual severity scores.
Offensive Security Becomes More Scalable
Traditionally, penetration testing has been constrained by cost, scope, and the availability of specialized professionals. Organizations often test only selected parts of their infrastructure, leaving other areas without regular assessment.
AI could fundamentally change this model by making automated testing more accessible and frequent. Eschelbeck believes organizations must also examine the entire chain between development and production, since even small code changes can have immediate consequences for production environments.
“The ability to do that, I think that’s really what AI made possible today,” he said. “I mean, AI is the tool to allow us to leverage penetration testing in an automated way and do it regularly, frequently, automatically, mostly autonomously, partially autonomously.”
For Terra Security, this shift reflects the growing need to understand what adversaries could accomplish across an organization’s changing attack surface, rather than simply identifying isolated vulnerabilities.
AI Could Help Level the Playing Field
Eschelbeck also discussed the impact of advanced AI models on the relationship between attackers and defenders. Models such as Mythos have attracted significant attention, partly because of their ability to identify security weaknesses and accelerate vulnerability discovery.
He pointed to the substantial wave of security patches released by organizations in recent months, arguing that AI is giving defenders access to capabilities that previously favored attackers.
“What I think those tools do, they bring balance into this inequality between the attacker and the defender. They really kind of neutralize the kind of advantage that the attackers had in the past.”
However, he also acknowledged that advanced models come with practical limitations, including cost and speed. Their impact will ultimately depend on how organizations integrate them into real security operations.
The Next Cybersecurity Giants May Be Startups
Perhaps one of Eschelbeck’s most interesting observations concerned where the next wave of cybersecurity innovation will come from. While established vendors remain influential, he believes the AI transition is creating opportunities for early-stage companies to build entirely new categories of security technology.
“I’m a big believer in the startup scene,” he said. “I think the biggest opportunity we have is following some of the early innovators in the space, looking at what is happening on the ground right now in the startup ecosystem.”
When Peled asked whether the next generation of major cybersecurity companies might emerge from today’s startups rather than existing players, Eschelbeck agreed: “That’s exactly right.”
AI has already created new security requirements around agents, models, infrastructure, and applications. These emerging challenges could create opportunities for companies building solutions specifically for the AI era.
A New Generation of Security Technology
The conversation highlighted how several parts of cybersecurity are evolving simultaneously, from vulnerability management and penetration testing to AI governance and agent oversight. As organizations adopt new technologies, they will need better ways to understand their exposure and test whether existing defenses can withstand real attacks.
For Eschelbeck, the next generation of cybersecurity innovation will come from companies willing to tackle these emerging problems. The opportunity for startups is not simply to automate existing processes, but to rethink how security works in an environment where both technology and threats are changing continuously.

