By Santosh Pai, Practice Head – IAM, Inspira Enterprise
We are constantly operating in an increasingly hyper-connected world, where traditional perimeters and defensive boundaries are no longer sufficient for enterprises. With organizations relying on cloud-based network services, hybrid work models, IoT devices, and AI-powered applications, the attack surface is expanding, and the traditional perimeter-based security model is getting obsolete. Enterprises are finding it increasingly difficult to consistently protect confidential data while staying ahead of evolving cyber threats.
Challenges with perimeter-based security
The traditional approach to network security, when a boundary is established separating the organization’s internal network from the external internet, is perimeter-based security. Firewalls, intrusion detection and prevention systems, among other measures, safeguard the perimeter, preventing unauthorized access to internal assets. However, in recent years, perimeter-based security has become outdated with the increased adoption of cloud technology, remote and hybrid work, BYOD culture, supply-chain risks, and threats posed by insiders.
When attackers can bypass the firewall, they are able to laterally move across the network, gaining access to sensitive data. Additionally, maintaining and upgrading perimeter infrastructure requires high and ongoing investments. These challenges with the perimeter-based model in securing modern digital infrastructure are compelling enterprises to shift to Zero-Trust architecture, which is based on the principle “never trust, always verify”.
Redefining trust in a borderless world
The perimeter-based model, which assumes that everything within the network can be trusted, is a misconception, as today’s attacks can originate from anywhere, including within the network itself. This is where Zero Trust Security can be implemented to address such challenges by following a pragmatic approach to cybersecurity, where no entity, internal or external, is trusted by default, but is subjected to validation. Zero Trust treats every access attempt as potentially malicious, irrespective of its origin. By integrating Zero Trust architecture, which assumes compromise is inevitable, with a cyber resilience strategy that focuses on business continuity, enterprises are protected continuously, with no interruption in business operations.
However, with the increase in agentic AI systems, where autonomous AI agents, which are active tools initiating actions, making decisions, and acting as non-human identities, can become a challenge to traditional identity and access management models.
Enabling Cyber Resilience through the Implementation of Zero Trust Architecture
According to the latest projections from Gartner, end-user spending on information security in India is expected to reach USD 3.4 billion in 2026, a 11.7% growth over the previous year. Despite this significant growth, the data breaches are reaching a record high every year, compelling organizations to prepare for, respond to, recover from, and adapt to cyber incidents without disrupting the business operations and ensuring organizations remain cyber resilient. This can be achieved through Zero Trust models. Zero Trust is the strategic approach built on key principles that collectively enhance cyber resilience. Nonetheless, with the agentic AI on the rise, these principles must evolve to address the AI agent. This identity scales rapidly and operates uninterruptedly, and interacts with speed, with governance adding to the complexity.
Continuous verification:
Organizations should adopt zero-trust security, as this approach functions by not trusting any entity, unlike perimeter-based security, where everyone once inside the network is implicitly trusted. This ensures every user, device, or application is verified, irrespective of their location, every time they attempt to access the network. Furthermore, the verification process must be continuous and not a one-off event, and must evolve to understand any deviation in AI agent behavior. This principle reduces the risk of malicious insiders or external attackers accessing and misusing sensitive data by continuously verifying every user action and request.
Identity and Access Management
In the Zero Trust model, user identities are authenticated, and access is provided only to approved users. This must extend to AI agents as well, in addition to human and device identities, and it should be noted that machine identities require stronger authentication mechanisms. Zero Trust model also leverages an authorization approach, which includes different levels of permission allowed to approved users. This ensures the right user is accessing the right level of information for the approved period of time. Some of the Identity and Access Management (IAM) systems are Multi-Factor Authentication (MFA), Single Sign-On (SSO), Privileged Access Management (PAM), and Identity Governance and Administration (IGA), among others. By implementing IAM solutions, organizations can reduce security risks significantly and enhance user experience by eliminating password fatigue.
Least Privilege Access
Least privilege access is a core component of the Zero Trust approach, where users and systems are granted only the minimum level of access necessary to perform their functions. With AI agents dynamically requesting access to perform tasks, ongoing validation and adaptive access controls should be implemented. The restriction reduces the risk of lateral movement, thereby limiting the scope of any potential damage the attacker intends to achieve and minimizing the attack surface. With the risk of a breach getting reduced, so is the minimization of system downtime, enhancing operational performance.
Microsegmentation
Here, networks are divided into isolated, granular, secure zones with specific access controls, limiting the blast radius of potential breaches and preventing attackers from moving freely across the systems. The barrier formed around the attacked zone could include a firewall or filter, limiting threats from moving out of the zone, while securing the rest of the network and strengthening cyber resilience. The risk of other segments getting breached is significantly reduced with microsegmentation. Furthermore, segmentation must be designed to accommodate AI-driven interactions across systems to ensure the AI agents are active only within these controlled boundaries.
In today’s digital era, where threats are inevitable, cyber resilience is the ultimate goal. Zero Trust provides a robust and scalable foundation to achieve this, enabling enterprises to operate securely in an increasingly complex digital landscape. By extending Zero Trust principles to include agentic AI, organizations are able to address the complexities the new technology brings. For modern enterprises, adopting Zero Trust is no longer optional; it is imperative.