Certified AI Security Professional (CAISP): Securing the Systems Everyone Else Is Building

Since the proliferation of generative AI and Large Language Models, AI Security has been treated as a subset of either AI development or traditional cybersecurity. It belonged to whoever happened to be closest to the problem. That arrangement no longer holds. AI systems have their own attack surfaces, failure modes, and threat landscape, and securing them requires expertise that neither a general security background nor a general AI background fully covers on its own. The Certified AI Security Professional (CAISP) credential exists to close that gap. 

CAISP equips cybersecurity and technology professionals with the knowledge and practical skills required to secure AI systems across their entire lifecycle. It treats AI security not as an afterthought bolted onto a finished model, but as a discipline integrated from design through deployment and monitoring. PECB is launching the credential at its annual conference in Rome, with an online phase on 21 and 22 September, followed by the in-person phase on 5 and 6 October. 

A Credential for Security and Technology Professionals

CAISP is designed for the people responsible for keeping AI systems trustworthy. The course is recommended for cybersecurity professionals, security architects, risk and compliance managers, AI and machine learning engineers, technology leaders, CISOs, and governance professionals. It is built for both the practitioner securing systems directly and the leader accountable for AI security across an organization. 

The course is structured to be accessible to experienced security professionals who are new to AI security, opening with the fundamentals of AI systems, machine learning, and large language models before progressing to advanced technical content. Participants with a general cybersecurity background and a basic awareness of how AI systems function will be well positioned to engage with the full course. 

What the CAISP Course Covers

The CAISP course goes beyond threat awareness to equip participants with the practical skills needed to defend and operate AI systems securely. This includes designing and deploying AI-specific security controls such as input validation, output filtering, and AI firewalls; integrating AI telemetry into SIEM platforms for monitoring and threat hunting; automating threat intelligence workflows using AI-powered tools; and conducting structured AI red team assessments. Participants also develop competencies in AI-driven incident response, supply chain governance, secure AI system architecture using zero trust principles, and compliance with emerging AI regulations including the EU AI Act. Hands-on lab exercises are embedded throughout the course to reinforce each of these areas in a practical, applied context. From there, it turns to the evolving AI threat landscape, which is where the credential does its most distinctive work.  

Participants gain hands-on experience identifying and mitigating threats that are specific to AI systems and largely absent from traditional security training: 

• Prompt injection attacks 

• Training data poisoning 

• Model inversion and data leakage 

• Model extraction and intellectual property risks 

• Adversarial input manipulation 

• AI supply chain vulnerabilities 

• Agent-based system risks 

These are not theoretical concerns. Each represents a category of attack that organizations deploying AI are already exposed to, and each requires a different defensive approach from the ones security teams are accustomed to applying. 

Why AI Security Matters Now

The case for AI security is no longer about anticipating a future problem. AI is in production across industries, which means the attack surface is already live. An AI model is a new kind of asset: it can be manipulated through its inputs, poisoned through its training data, reverse-engineered to expose the data it was trained on, or extracted wholesale as stolen intellectual property. The systems that organizations are racing to deploy are, in many cases, being secured by teams who have not yet been trained in how these attacks work. 

CAISP addresses this directly by combining awareness of AI-specific attack surfaces, the ability to assess AI security risk with confidence, and the capability to implement controls aligned with emerging standards. It positions the credential holder to protect intellectual property and sensitive data, and to lead AI security initiatives rather than react to incidents after they occur. For a security professional, this is the difference between defending AI systems competently and defending them at all. 

A Step Worth Taking 

Whether you are a security professional moving into AI, an AI engineer taking on security responsibility, or a leader accountable for both, CAISP offers a structured way to demonstrate that you can secure AI systems to a recognized standard. As AI security becomes a discipline, certified expertise distinguishes the professionals who can be trusted with it.  

CAISP is being launched at the PECB Conference in Rome and will be delivered by Graeme Parker, an experienced cybersecurity practitioner and ISO/IEC 27001 expert, and will award 31 CPD credits on completion.

Learn more about the credential and reserve a place: