Webinar Description
Key Takeaways
- Explores the shift from human-centric identity frameworks to machine-first approaches for autonomous AI agents
- Introduces cryptographic SPIFFE-based universal workload identity as a foundational trust layer
- Covers OAuth-based token flows for agent authorisation in both autonomous and delegated scenarios
- Addresses least privilege enforcement and governance controls operating at machine speed
- Relevant for security leaders, IAM specialists and IT professionals deploying autonomous agents
Introduction
As artificial intelligence agents transition from simple assistants to fully autonomous entities capable of independent decision-making, organisations face a fundamental security challenge: how to authenticate, authorise and govern machine identities operating at speeds that exceed human oversight. The webinar “Securing the Agentic Frontier: Machine-Scale Identity and Authorization for Autonomous Agents” addresses this emerging concern, presenting a technical blueprint for identity security in agentic environments. Hosted by CyberArk, a Palo Alto Networks company, the session targets security leaders, AI practitioners and identity and access management professionals who are navigating the complexities of deploying autonomous agents within enterprise infrastructure.
About This Event
This live virtual webinar examines the security architecture required when AI agents operate independently rather than under direct human control. The session moves beyond theoretical discussion to provide actionable strategies for implementing cryptographic agent identities, establishing trust relationships between autonomous systems, and applying governance controls that function effectively at machine scale. The programme draws on the combined expertise of CyberArk and Palo Alto Networks, integrating their approaches to workload identity and AI agent security.
The Challenge of Machine-Scale Identity
Traditional identity and access management frameworks were designed around human users—individuals who authenticate periodically, request access to specific resources, and operate within predictable patterns. Autonomous AI agents fundamentally disrupt this model. These systems may initiate thousands of actions per second, interact with multiple services simultaneously, and make decisions without human intervention. The identity frameworks that served organisations well for human workforces cannot simply be extended to accommodate this new class of non-human actors.
The webinar addresses this gap by advocating for a deliberate shift toward machine-first identity architectures. Rather than treating agent identities as an afterthought or attempting to retrofit human-centric controls, the session presents an approach that recognises autonomous agents as first-class entities requiring purpose-built security mechanisms. This reframing has significant implications for how organisations design authentication flows, establish trust boundaries, and implement access controls.
SPIFFE as a Foundational Trust Layer
Central to the webinar’s technical framework is SPIFFE, the Secure Production Identity Framework for Everyone. SPIFFE provides a standardised method for issuing cryptographic identities to workloads—software processes running in diverse environments including containers, virtual machines and serverless functions. By extending SPIFFE principles to autonomous agents, organisations can establish strong, verifiable identities that do not depend on shared secrets or static credentials.
The cryptographic foundation that SPIFFE provides addresses several persistent challenges in agent security. Identities become portable across platforms, enabling agents to authenticate consistently whether they operate in cloud environments, on-premises infrastructure, or hybrid deployments. The framework also supports short-lived credentials that reduce the risk associated with credential theft, a particularly important consideration when agents may interact with sensitive systems without human oversight.
Interoperability represents another significant advantage. As organisations deploy agents from multiple vendors or develop custom autonomous systems, a universal identity standard prevents the fragmentation that would otherwise occur if each platform implemented proprietary authentication mechanisms. SPIFFE-based identities can serve as a common trust anchor across heterogeneous agentic environments.
OAuth-Based Authorisation for Autonomous and Delegated Agents
While SPIFFE addresses the question of identity—establishing who or what an agent is—authorisation determines what that agent may do. The webinar explores how OAuth-based token flows can govern agent permissions, accommodating both fully autonomous operation and scenarios where agents act on behalf of human users.
This distinction matters considerably in practice. An autonomous agent performing infrastructure monitoring may require standing permissions to access system metrics, while an agent executing tasks delegated by a specific user should inherit only the permissions appropriate to that user’s role. OAuth’s token-based architecture provides the flexibility to model both scenarios, with tokens encoding the scope of permitted actions and the context under which they were granted.
The session examines how Idira Secure AI Agents by Palo Alto Networks integrates SPIFFE-based identity with OAuth authorisation, creating a layered security model. This integration ensures that strong authentication precedes any authorisation decision, and that tokens issued to agents carry verifiable claims about the agent’s identity and the permissions it has been granted.
Enforcing Least Agency and Governance at Machine Speed
The principle of least privilege—granting only the minimum permissions necessary to perform a task—has long been a cornerstone of security practice. When applied to autonomous agents, this principle evolves into what the webinar terms “least agency,” recognising that agents possess not merely access rights but the capacity for independent action. Constraining agency requires controls that limit not only what resources an agent can access but what actions it can initiate and under what conditions.
Governance presents particular difficulties when agents operate at machine speed. Human review of individual agent actions becomes impractical when those actions occur in milliseconds. The webinar addresses this challenge by exploring automated governance mechanisms that can evaluate agent behaviour against policy in real time, intervening when actions exceed defined boundaries without introducing latency that would negate the efficiency benefits of autonomous operation.
Who Should Attend
The webinar is designed for professionals responsible for securing AI deployments and managing non-human identities within enterprise environments. Security leaders evaluating the risks associated with autonomous agents will find the session’s framework useful for developing governance strategies. Identity and access management specialists can expect detailed technical discussion of how existing IAM architectures must evolve to accommodate machine identities. AI practitioners and technical managers involved in deploying agentic systems will gain insight into the security requirements that should inform their implementation decisions.
Organisations at various stages of AI agent adoption may benefit from the session. Those in early planning phases can use the blueprint presented to establish security foundations before deployment, while enterprises with existing agent deployments may identify gaps in their current approaches and opportunities to strengthen their security posture.
Industry Context
The timing of this webinar reflects broader industry developments. As large language models and other AI technologies mature, organisations are increasingly deploying agents that can perform complex tasks with minimal human supervision. This expansion of autonomous capability has outpaced the development of corresponding security frameworks, creating a gap that threat actors may exploit. Regulatory attention to AI governance is also intensifying, with emerging requirements around accountability and transparency that depend on robust identity and audit mechanisms.
The convergence of identity security and AI represents a significant evolution for both disciplines. Identity professionals must understand the unique characteristics of machine actors, while AI teams must recognise that security cannot be an afterthought in agentic architectures. This webinar sits at that intersection, offering a perspective informed by expertise in both domains.