phone iconSpeak with our expert +14109365774

Trusted by millions of learners

Learn more about the course

Get details on syllabus, projects, tools, and more

Name
Email
Mobile Number

By submitting this form, you consent to our Terms of Use & Privacy Policy and consent to be contacted via email, phone (including by AI-generated/pre-recorded voice calls), SMS, or WhatsApp.

Certificate Program in AI Security

Certificate Program in AI Security

Application closes 24th Nov 2026

Why should you join this program?

  • List icon

    Flexible Learning Paths

    This program offers learners the flexibility to learn with or without code, featuring dynamic variations of activities suited for technology professionals and strategy leaders alike.

  • List icon

    Learn from JHU, a Leading US Research University

    Ranked #2 in Computer Information Technology, #9 National University, and #17 Best Global University, reflecting JHU's leadership in research and innovation. (2027 Rankings)

LEARNING OUTCOMES

What will you learn to build and apply?

Through a structured learning journey, you will build the capability to:

  • Analyze the AI threat landscape and design secure architectures to resist misuse and multimodal attacks

  • Understand AI security frameworks and address prompt injection, jailbreaks, adversarial inputs and model abuse

  • Understand AI data pipeline security through data lineage, privacy controls, integrity, and attack mitigation

  • Understand alignment techniques, trust boundaries, tool abuse, and multi-agent risks in agentic AI systems

  • Assess AI models for safety, fairness, and robustness, and explore runtime guardrails and jailbreak testing

  • Explore AI governance for secure delivery, compliance monitoring, human oversight, and regulatory requirements

Earn a Certificate of Completion from Johns Hopkins University

  • #9 National University Rankings

    #9 National University Rankings

    U.S. News & World Report, 2027

  • #2 Computer Information Technology

    #2 Computer Information Technology

    U.S. News & World Report, 2027

  • #11 Most Innovative Schools

    #11 Most Innovative Schools

    US News and World Report, 2027

  • #17 Best Global University

    #17 Best Global University

    U.S. News & World Report, 2027

  • #1 Biomedical Engineering Program

    #1 Biomedical Engineering Program

    US News and World Report, 2027

KEY PROGRAM HIGHLIGHTS

Why choose this program?

  • List icon

    Learn from Johns Hopkins University Faculty

    Learn through recorded lectures and attend monthly faculty-led masterclasses to secure AI systems across their entire lifecycle.

  • List icon

    Interactive Mentorship by Industry Experts

    Learn from experts through weekly live mentorship sessions focused on industry best practices to secure AI and Agentic AI systems.

  • List icon

    Comprehensive Curriculum

    Learn from an in-depth curriculum that covers security across the entire AI development lifecycle, from data and model design through deployment and monitoring.

  • List icon

    Dynamic Hands-On Activities

    Learn to secure modern AI agents and LLM-based architectures through dynamic code, no-code, and low-code projects, lab activities, tools, and technologies.

  • List icon

    Earn a Recognized Credential from Johns Hopkins University

    Earn a Certificate of Completion and 6.3 CEUs from Johns Hopkins University upon successful completion of the program.

  • List icon

    Personalized Program Support

    Get personalized assistance from a dedicated Program Support Team and AI-assisted learning support through Great Learning.

Skills you will learn

Agentic AI Security

Runtime Guardrail Integration

Tool Abuse Testing

Guardrail Implementation

AI Threat Modeling

Automated Agentic Workflow

AI Governance & Compliance

Model Card Analysis

AI Security Forensics

Multi-Agent Trust Boundary & Message Verification

Agentic AI Security

Runtime Guardrail Integration

Tool Abuse Testing

Guardrail Implementation

AI Threat Modeling

Automated Agentic Workflow

AI Governance & Compliance

Model Card Analysis

AI Security Forensics

Multi-Agent Trust Boundary & Message Verification

view more

  • Overview
  • Learning Journey
  • Curriculum
  • Projects
  • Tools
  • Certificate
  • Faculty
  • Mentors
  • Fees
  • FAQ

Who is the program for?

This program is exclusive to learners in eligible JHU AI Certificate Programs, in roles such as:

  • AI and Data

    Professionals building or deploying AI solutions who want to develop system-level risk awareness and skills to secure AI across its lifecycle.

  • Cybersecurity and Tech

    Professionals who want to expand their security expertise to LLMs, RAG architectures, AI agents, model supply chains, and AI workflows.

  • Leadership and Governance

    Professionals who need to govern risk, ensure compliance, and evaluate secure architectures without writing code.

How is the program learning experience?

Our pedagogy is designed to ensure a holistic learning experience

  • banner-image

    Learn from Experts

    Learn from JHU faculty and industry experts to secure AI and Agentic AI systems

  • banner-image

    Learn By Doing

    Apply Agentic AI concepts through hands-on projects and real-world case studies

  • banner-image

    Earn a University Credential

    Earn a certificate of completion and 6.3 CEUs from Johns Hopkins University

  • banner-image

    Get Support Throughout the Learning Journey

    Program managers will help you stay on track, navigate key milestones & complete the program

Comprehensive curriculum

The curriculum for the Certificate Program in AI Security, crafted by expert faculty at Johns Hopkins University and leading industry experts, covers security across the AI development lifecycle, from data and model design through deployment and monitoring, equipping you with the skills to secure AI systems at every stage.

  • Code & No-Code

    Tools and Techniques

  • Dynamic

    Projects and Activities

  • Monthly

    Faculty-Led Masterclasses

PREPARATORY MODULE

This preparatory module ensures that learners have the technical foundations required to begin the program. They will build proficiency in Python programming, understand the fundamentals of AI systems, and gain foundational knowledge of cybersecurity principles.

MODULE 01: SECURE AI DESIGN AND THREAT LANDSCAPE

In this module, learners will build a strong foundation in the different types of attacks that target AI systems. Learners will be introduced to the CIA Triad, which is the DNA of cybersecurity. They will explore different approaches to building secure architecture for AI applications, and examine why strict trust boundaries and airlocks are essential. Learners will also shift from reactive patching to privacy by design, containing the blast radius early in the AI lifecycle.

WEEK 01: AI SECURITY BY DESIGN

In this week, learners will explore the core threat landscape, from foundational CIA Triad principles to complex multi-model and agent architectures. They will examine how to identify architectural flaws before deployment, map security boundaries for RAG, and structure multi-model workflows. Applying threat modeling and security boundaries ensures that lower-cost and frontier models collaborate safely without expanding the attack surface. Topics Covered: - Introduction to AI Threats and Vulnerabilities - Confidentiality, Integrity, and Availability of the Model Design; Detecting and Preventing Model Misuse and Abuse - Threat Modeling - Secure RAG Architecture - Agent Security Architecture - Secure LLM Architecture - Multi-Model AI Systems: Security Risks in Multi-Model AI Systems, Multimodality, Cost Optimization, and Security Frameworks for Combining Lower and Frontier Models

MODULE 02: SECURING THE AI BUILD LIFECYCLE

In this module, learners will secure the data pipeline, protecting the crown jewels of an AI system against hidden backdoors and compromised training inputs. They will implement these controls to build systemic resilience across every stage of development and deployment, moving step-by-step through the hands-on engineering phase of the AI lifecycle. Learners will also verify data sources, precisely align model behavior, and deploy runtime guardrails to keep AI applications safe, compliant, and under control.

WEEK 02: DATA INTEGRITY AND SUPPLY CHAIN SECURITY

In this week, learners will explore how data pipelines feed the core reasoning engine, making them a crown jewel of AI systems. They will practice how to spot poisoned training data, prevent hidden backdoor vulnerabilities, trace data lineage, and audit third-party AI supply chains so the foundation remains trustworthy. Topics Covered: - Data Processing and Preparation - Advanced Data Storage and Retrieval - Securing the Data Pipeline - Data Privacy and Safety Controls - Data-Specific Attacks: Data Poisoning - Backdoor Attacks via Training Data - Data Lineage - ML Supply Chain

WEEK 03: MODEL EDITING – ABLATION, ALIGNMENT, AND SUPERVISED FINE-TUNING

In this week, learners will explore different ways of implementing mechanisms of model misuse and preventing alignment failures. They will be equipped with the skills to debug models, correct built-in biases, and apply fine-tuning techniques like RLHF (Reinforcement Learning from Human Feedback). Model editing through ablation and alignment gives defenders the precision to remove embedded biases and enforce safety guardrails, without sacrificing performance or erasing what the model has already learned. Topics Covered: - Ablation: From Neuron Lesioning to Component Isolation - Ablation for Security Debugging and Bias Correction - Recovery and Verification: Distillation and Before/After Evaluation (Ablation or De-biasing) - Alignment by Design: Specs, Refusal Behavior, and the Helpful/Harmless Frontier - De-Alignment, Cornering, and Sovereignty - RLHF and Supervised Fine-Tuning in Practice: Curation, Mechanics, and Regression Evaluations

WEEK 04: LEARNING BREAK

WEEK 05: SECURE DEVELOPMENT OF MODELS AND AGENTS

In this week, learners will focus on pre-release verification to ensure system components interact safely before reaching production users. They will explore AI evaluations across safety, quality, and performance dimensions, including responsible AI assessments for bias, fairness, and harmful content. The week also covers securing model training and operating environments as part of a rigorous build lifecycle. By the end, learners will be equipped to validate agent interactions, measure model robustness against adversarial prompts, and ensure reliable system behavior under real-world conditions. Topics Covered: - Securing Model Training and Operating Environments - AI Evaluations Across Safety, Quality, and Performance Dimensions; AI Engineering - Responsible AI Assessments for Bias, Fairness, and Harmful Content - Model Robustness Testing Against Hallucinations and Reasoning Failures

WEEK 06: MODEL RESILIENCE AND GUARDRAILS

In this week, learners will explore how placing an AI model into production exposes it to live adversarial traffic, where static security filters quickly fall short. Implementing dynamic runtime guardrails protects the system in production, helping prioritize the specific real-world ways attackers are exploiting these models. Learners will apply these guardrails, from jailbreak assessments and rate limiting to automated output verifiers, which intercept live prompt injections, block unauthorized agent actions, and keep AI systems in production from being exploited or misused. Topics Covered: - Comprehensive Security Testing for AI Agents, RAG Systems, and LLMs - Jailbreak and Prompt Injection Assessments - Model Security Reviews - Defenses Against Model Theft and Extraction Attacks - Secure Deployment With Policy Engines and Schema Constraints - Runtime Guardrails: Output Validation and Response Verifiers - Security Controls and Guardrails for AI Systems - Defensive Measures: Rate Limits, Canaries, Watermarking, Query Auditing

WEEK 07: LEARNING BREAK

MODULE 03: DEPLOYMENT, OPERATIONS, AND GOVERNANCE

In this module, learners will unpack what it truly means to own a model in production, where performance can drift, integrations can crack, and compliance rules shift without warning. Continuous governance transforms compliance from a static checklist into an active, scalable defense against performance drift and operational risk. From monitoring real-time model behavior to managing third-party risk and building human oversight that scales, this module prepares learners to keep enterprise AI trustworthy long after it goes live.

WEEK 08: INFRASTRUCTURE SECURITY AND AI SECURITY OPERATIONS

In this week, learners will explore how operating AI in production demands purpose-built pipelines that go beyond traditional DevOps. They will implement Artificial Intelligence Security Operations, integrating cryptographic verification, secure model registries, AI Software Bill of Materials (SBOM), secret management, and agent sandboxing into every deployment cycle. Learners will also gain the skills to automate safety gates, enable instant rollback capabilities, and isolate autonomous tools so vulnerabilities are contained before they escalate into operational risk. Topics Covered: - AISecOps: CI/CD for AI Systems, Deployment Gates, Rollback - Securing the Model Registry: Versioning, Signoff, Approvals - Artifact Integrity: Signing, Attestation - Secret Management for AI Systems - AI Software Supply Chain Security - AI SBOM (Software Bill of Materials) - Vector Database Security (Operational View) - Agent Sandboxing and Tool-Abuse Prevention - Comprehensive Security Testing on AI Agents, RAG, LLMs

WEEK 09: AI GOVERNANCE AND MONITORING

In this week, learners will examine how slow, hidden threats can quietly erode AI system integrity over time. Implementing continuous observability frameworks ensures autonomous machine behavior is audited at scale before incidents escalate. Learners will also develop the judgment to know when to step in, building real-time monitoring habits, protecting sensitive data, and responding to incidents with the confidence and clarity that keep AI systems running safely. Topics Covered: - AI Governance - Governance for AI Agents - Model Monitoring (Drift, Data Quality) - AI Governance Maturity Metrics - Observability for Agentic Systems - Incident Response for AI Systems - Data Leak Prevention - Human-in-the-Loop Design: Where, When, How

SELF-PACED MODULE: CLAUDE-BASED AI WORKFLOWS

This module is designed to build practical capability in applying Generative AI and Agentic AI using the Claude ecosystem in real-world contexts. Participants build the ability to design, execute, and evaluate AI-driven workflows for real-world applications, supported by structured learning. *Disclaimer: Access to tools within the Claude ecosystem is not included as part of the program. Participants may choose to explore advanced capabilities independently.

Design and Execute AI Workflows

- Model selection and prompt engineering using Claude Chat - Agentic workflow design and orchestration using Claude CoWork - Plan → Approve → Execute → Iterate framework - Designing workflows with reasoning, tools, and multi-step execution - Applying concepts through real-world case studies

Build and Deploy AI Systems at Scale

- API integration and model usage using Claude Code - Tool integration using the Model Context Protocol - Designing agentic systems with memory, tools, and orchestration - Performance optimization, cost considerations, and system reliability - Responsible AI principles, including alignment approaches such as Constitutional AI

MASTERCLASSES

Learn through monthly live masterclasses from Johns Hopkins University faculty

ATTACK ON AI

In this masterclass, learners will be placed at the front lines of real-world AI attacks, studying how today's most damaging breaches unfolded, what warning signs were missed, and what it cost. Through MITRE ATLAS and OWASP frameworks, learners will dissect live threat patterns to ensure history does not repeat itself. By the end, learners will be able to evaluate emerging threats and weigh AI risk against business value. Topics Covered: - Industry Case Studies (Data Breaches, Statistics) - MITRE ATLAS - OWASP Top 10 for LLMs - Strategic Value vs. Risk Mapping for AI Use Cases - Why AI Security Is Different From Traditional InfoSec - Tech Leadership Considerations: Where AI Security Investment Pays Off - Privacy by Design

SECURE ARCHITECTURE AND DEPLOYMENT OF AI AGENTS

In this masterclass, learners will explore why giving autonomous agents direct execution power introduces risks that standard security controls simply were not built to handle, especially as the stakes change entirely once AI agents can act on their own. They will discover how separating what an agent thinks from what it can do is the critical design choice that stops failures from cascading across multi-agent systems. Topics Covered: - Agentic AI Types - Agentic AI Workflow: Orchestrated Workflow → Single Agent → Multi-Agent → Autonomous - Model Context Protocol (MCP): Trust Boundaries, Client-Host-Tool Chain - MCP Attack Vectors - Security Use Cases of Agentic AI - Incident Response for Agentic Systems - Monthly Frontier Briefing: Current Incidents, Current Mitigations

ANTHROPIC

In this masterclass, learners will explore the Anthropic AI landscape, covering Claude models, Constitutional AI, and key safety and alignment principles. They will apply effective prompt engineering techniques, use the Claude API for tasks and integrations, generate structured outputs, build simple applications, compare Claude with other AI models, and evaluate ethical considerations for deploying AI systems.

SAMPLE CODE-BASED PROJECTS

Note: This is an indicative list of projects and is subject to change.

BACKDOOR IMPLANTATION VIA DATA POISONING: DETECTION, REMOVAL, AND INCIDENT RESPONSE

Domain: AI Security / Adversarial ML / Incident Response Objective: Implant a backdoor into a machine learning model through data poisoning, then apply detection and removal techniques to cleanse the model. Document the full lifecycle in an incident report covering root cause, indicators of compromise, and recommended pipeline mitigations. Skills You Will Learn: Data Poisoning, Backdoor Detection, Model Cleansing, Incident Response, Adversarial ML Defense, AI Supply Chain Security

ABLATION AND DISTILLATION ON A TABULAR FRAUD-DETECTION MODEL FOR SECURITY AND BIAS REMEDIATION

Domain: AI Security / Model Robustness / Responsible AI Objective: Apply targeted ablations to a fraud-detection model to address a security vulnerability and a bias issue, then use knowledge distillation to recover lost performance. Validate results through before-and-after evaluation of accuracy, robustness, and fairness. Skills You Will Learn: Ablation Techniques, Tabular ML Security, Bias and Fairness Analysis, Knowledge Distillation, Responsible AI Assessment

EXPLORING TOOL ABUSE SCENARIOS IN AGENTIC AI AND RECORDING RESULTS

Domain: AI Security / Agentic AI Security / Red Teaming Objective: Test and evaluate how an agentic AI system's tools and capabilities can be manipulated through adversarial inputs to perform unintended actions. Record findings in a structured log and propose guardrails and least-privilege controls to mitigate identified risks. Skills You Will Learn: Agentic AI Red Teaming, Prompt Injection, Tool Abuse Testing, Trust Boundary Analysis, Security Logging, Guardrail Design

SAMPLE NO-CODE / LOW-CODE PROJECTS

Note: This is an indicative list of projects and is subject to change.

BUILDING AND SECURING AN AI WORKFLOW USING A LOW-CODE PLATFORM

Domain: AI Security / Workflow Automation Security Objective: Design and implement an automated AI workflow using a low-code visual platform, connecting AI models with external tools and data sources. Identify and address common security risks in automated workflows such as insecure connections, over-permissive integrations, and unvalidated inputs. Skills You Will Learn: Workflow Automation, AI Integration, Secure Configuration, API Security Basics, Credential Management

GENERATING AN AI BILL OF MATERIALS (AI-BOM) USING NO-CODE TOOLS

Domain: AI Supply Chain Security Objective: Use no-code tooling to catalog the key components of an AI-enabled application, including models, datasets, libraries, and third-party dependencies. Produce a structured inventory that supports transparency, license tracking, and supply chain risk management. Skills You Will Learn: AI-BOM Generation, Component Inventory, Supply Chain Security, Provenance Tracking, License Management

MANAGING SUPPLY CHAIN RISKS BY ANALYZING MODEL CARDS

Domain: AI Supply Chain Security / AI Governance Objective: Evaluate publicly available model cards to assess risks related to training data, intended use, bias, and licensing before integrating third-party models into a production environment. Deliver a structured risk assessment and governance checklist based on findings. Skills You Will Learn: Model Card Analysis, Supply Chain Risk Assessment, Model Provenance Evaluation, Third-Party Model Governance

Note: The curriculum listed above is indicative and subject to updates as technology evolves.

What activities will you learn from?

Build hands-on AI security skills through immersive labs, and real-world case studies.

project icon

SAMPLE LAB ACTIVITY

Threat Modeling and Attack Testing on an LLM Application

Description

Build a threat model for a provided LLM-based application by mapping its attack surface and data flows. Design and execute a set of prompt injection and jailbreak attacks to validate identified risks, referencing established LLM security frameworks.

Skills you will learn

  • LLM Threat Modeling
  • Prompt Injection Testing
  • Attack Surface Mapping
  • Risk Assessment
project icon

SAMPLE LAB ACTIVITY

Implementing Runtime Guardrails on an LLM Application

Description

Apply runtime guardrails to a baseline LLM application to enforce topic boundaries, block adversarial inputs, and prevent data leakage. Run a simulated incident to test detection and response capabilities.

Skills you will learn

  • NVIDIA NeMo Guardrails
  • Guardrail Implementation
  • Input/Output Validation
  • AI Monitoring
  • Incident Simulation
project icon

SAMPLE LAB ACTIVITY

Building an AI Compliance Mapping Matrix

Description

Construct a comparative matrix mapping major AI regulations and frameworks against common control areas such as risk, transparency, and accountability. Use the matrix to identify compliance gaps and overlaps relevant to real-world AI deployments.

Skills you will learn

  • AI Governance and Compliance (GRC)
  • Regulatory Analysis
  • Control Mapping
  • Compliance Assessment
project icon

SAMPLE LAB ACTIVITY

Prompt Airlines: AI Security CTF Challenge

Description

Participate in a guided Capture the Flag challenge focused on interacting with and exploiting LLM-based applications. Practice identifying vulnerabilities and applying adversarial thinking in a safe, gamified environment.

Skills you will learn

  • Prompt Injection
  • Jailbreak Techniques
  • LLM Red Teaming
  • Guardrail Bypass Analysis
project icon

SAMPLE CASE STUDY

Claude Mythos: When AI Becomes the World's Best Hacker

Description

Explore how Anthropic's Claude Mythos AI independently found and exploited serious security flaws during controlled tests without human help, revealing how AI could become a powerful and dangerous hacking tool for attackers.

Skills you will learn

  • AI Threat Modeling
  • Cyber Risk Governance
  • AI Misuse Detection
  • Proactive Defense Planning
project icon

SAMPLE CASE STUDY

Inside OpenClaw: Real Attacks on Autonomous AI Agents

Description

Understand how hackers manipulate AI assistants like OpenClaw using hidden instructions, rogue plug-ins, and software weaknesses to silently steal passwords, files, and private data without the user ever knowing they were attacked.

Skills you will learn

  • AI Agent Security
  • Prompt Manipulation Defense
  • System Access Controls
  • Risk Assessment
project icon

SAMPLE CASE STUDY

First Fully Autonomous AI Attack

Description

Examine how OpenAI's AI models broke out of a controlled test on their own and hacked into Hugging Face's systems, the world's first fully AI-directed cyberattack, highlighting the urgent need for stronger AI containment.

Skills you will learn

  • Autonomous Attack Detection
  • System Containment
  • AI Risk Management
  • Security Forensics
project icon

SAMPLE CASE STUDY

Camoleak and Echoleak: Zero-Click Exploits on RAG and Agents

Description

Examine how attackers secretly stole private files, passwords, and corporate data from AI work tools like Microsoft 365 Copilot and GitHub Copilot just by sending a crafted email or hiding instructions in code with no user action needed.

Skills you will learn

  • RAG Security Testing
  • Zero-Click Exploit Analysis
  • Indirect Prompt Injection
  • LLM Scope Violation Detection
  • Secure Copilot Deployment
project icon

SAMPLE CASE STUDY

DeepSeek and the Cross-Border Data Security Crackdown

Description

Explore how governments worldwide responded to DeepSeek's AI launch through immediate bans, 13 European investigations, and corporate restrictions after hidden code was found secretly sending user data to Chinese state-linked servers, raising serious privacy and national security concerns.

Skills you will learn

  • AI Geopolitical Risk Analysis
  • Privacy Impact Assessment
  • Hidden Telemetry Auditing
  • AI Sovereignty and Vendor Risk Management
  • Regulatory Intelligence
  • Cross-Border Data Compliance

This is an indicative list of lab activities and case studies and is subject to change.

What no-code and code tools will you explore?

Work with a comprehensive suite of modern code-based or no-code tools

  • tools-icon

    Claude

  • tools-icon

    OpenAI

  • tools-icon

    Gemini Notebook

  • tools-icon

    Teachable Machine

  • tools-icon

    n8n

  • tools-icon

    KNIME

  • tools-icon

    Zapier

  • tools-icon

    Google Colab

  • tools-icon

    Python

  • tools-icon

    VS Code (Visual Studio Code)

  • tools-icon

    Jupyter Notebook

  • tools-icon

    OWASP Threat Dragon

  • tools-icon

    Hugging Face

  • tools-icon

    Vector Databases

  • tools-icon

    DSPy

  • tools-icon

    LangChain

  • tools-icon

    LlamaIndex

  • tools-icon

    Amazon CloudWatch

  • tools-icon

    Databricks

This is an indicative list, and relevant tools are reviewed for inclusion periodically. Access to certain tools may require additional subscriptions or fees.

Earn a Certificate of Completion from Johns Hopkins University

Earn a globally recognized credential from a top U.S. university and showcase it to your network.

certificate image

* Image for illustration only. Certificate subject to change.

Who are the faculty for the program?

Learn from renowned JHU faculty and build the intuition and strategic judgment to secure modern AI systems

  • Ian McCulloh, Ph.D.

    Ian McCulloh, Ph.D.

    Director, AI Executive & Professional Education, Johns Hopkins University

    Served as Chief Data Scientist & MD of AI, Accenture Federal Services

    Author of three books and over 100 peer-reviewed papers

    Know More
    Company Logo
  • Kiran Chittargi

    Kiran Chittargi

    Adjunct Faculty - Computer Science, Johns Hopkins University

    Program Director at Microsoft

    Global Expansion Lead for Azure

    Know More
    Company Logo
  • Dr. Pedro Rodriguez

    Dr. Pedro Rodriguez

    Faculty, Johns Hopkins University AI Program

    Leads the Information Science Branch at the JHU's Applied Physics Lab

    20+ years of expertise in AI/ML algorithms

    Know More
    Company Logo
  • Dr. Art Sedighi

    Dr. Art Sedighi

    Adjunct Faculty, Johns Hopkins University

    Ph.D. in Systems Engineering and in Cybersecurity/Cryptography

    AI and security expert focused on LLM security, & cloud architectures

    Know More
    Company Logo

Who are the mentors for weekly live sessions?

Learn from seasoned AI industry mentors to apply concepts and build practical skills

  •  Aniket Amdekar - Mentor

    Aniket Amdekar linkin icon

    General Manager - Cyber Defence Education Great Learning
    Great Learning Logo
  •  Ivan Marchany  - Mentor

    Ivan Marchany linkin icon

    Cyber Threat Engineer, Raymond James
    Company Logo

Note: This is an indicative list and is subject to change based on the availability of mentors.

Program Fee

The program fee is USD 1,800

Invest in your career

  • benifits-icon

    Build hands-on expertise in operationalizing responsible AI practices across the enterprise

  • benifits-icon

    Dedicate 8-9 hours weekly to faculty-led learning, industry mentorship, projects, and case studies

  • benifits-icon

    Learn from AI experts in weekly live online sessions focused on real-world case studies

  • benifits-icon

    Receive a Certificate of Completion and 6.3 CEUs from Johns Hopkins University

Take the next step

timer
00 : 00 : 00

Apply to the program now or schedule a call with a program advisor

Learn more about the program

Application closes: 24th Nov 2026

Application closes: 24th Nov 2026

Talk to our advisor for offers and course details

Application process

Applications close once the required number of participants enroll. Apply early to secure your spot

  • steps icon

    Fill application form

    Apply by filling out a simple online application form.

  • steps icon

    Review process

    A panel from Great Learning will review your application to determine your fit for the program.

  • steps icon

    Join program

    Receive an offer for a seat in the upcoming cohort of the program post a final review.

Exclusive to JHU AI Certificate Program learners

  • This program is exclusively available to learners enrolled in AI certificate programs offered in collaboration with Johns Hopkins University.
  • Contact your Program Advisor to understand which programs are eligible for enrollment.

Frequently Asked Questions

Program Details
Faculty, Curriculum and Projects
Eligibility & Enrollment
Fee & Payment
Career-related Queries
General Queries
Role-Based Queries
Program Details

What does the Johns Hopkins University Certificate Program in AI Security cover?

The nine-week online program covers AI threat modeling, secure AI architecture, data integrity, model alignment, adversarial testing, runtime guardrails, AI security operations, monitoring, and governance. It combines hands-on labs, real-world case studies, and practical projects to help participants develop skills in securing AI systems across their development lifecycle. Participants can choose code-based or no-code/low-code learning activities. Note: The curriculum topics listed above are indicative and subject to updates as technology evolves.

What certificate do participants receive after completing the program?

Participants who successfully complete the program receive a Certificate of Completion from Johns Hopkins University and 6.3 Continuing Education Units (CEUs). They also build a shareable e-portfolio showcasing their AI security skills and practical project work.

How is the JHU AI Security Program different from other AI security courses?

The Johns Hopkins University (JHU) Certificate Program in AI Security stands out for its features like:
●​ Flexible learning formats: Code-based and no-code/low-code options cater to technical professionals and business leaders.​
●​ Advanced AI security coverage: Topics include LLM security, MCP trust boundaries, model alignment, AI supply chain security, and agentic AI threats.​
●​ Structured mentorship: Participants benefit from weekly mentored learning sessions, monthly JHU faculty masterclasses, discussion forums, and AI-assisted learning support through Great Learning.​
●​ University credential and practical experience: Successful participants earn a Certificate of Completion and 6.3 CEUs from JHU, while building a shareable e-portfolio through hands-on projects and a capstone.

Who should consider this AI security program?

The program is designed for AI and data professionals who want to secure the systems they build, Cybersecurity and technology professionals who want to expand their expertise to AI-specific threats, and AI leadership and strategy professionals who need to manage AI risks, governance, and compliance.

Faculty, Curriculum and Projects

Who teaches the Johns Hopkins University AI Security program?

The program features Johns Hopkins University faculty and industry experts. The brochure identifies Dr. Ian McCulloh, Kiran Chittargi, Dr. Pedro Rodriguez, and Dr. Art Sedighi among the faculty. Participants also benefit from monthly faculty-led masterclasses and weekly mentored learning sessions with industry experts.

Does the program cover LLM security, prompt injection, and jailbreak attacks?

Yes. The program covers LLM security, prompt injection, jailbreak assessments, adversarial inputs, model extraction, and defenses against AI misuse. Participants also explore frameworks such as MITRE ATLAS and the OWASP Top 10 for LLMs to understand and assess AI-specific threats.

What projects can I work on in the AI security program?

The program includes code-based and no-code/low-code project options. Examples include detecting and removing backdoors introduced through data poisoning, using model ablation and distillation for security and bias remediation, testing tool-abuse scenarios in Agentic AI, and building and securing automated AI workflows. Note: The projects listed above are indicative and subject to updates to the curriculum.

What practical tools and frameworks are covered?

The program introduces a range of tools and technologies across its code-based and no-code/low-code learning formats. No-code tools: Claude, OpenAI API, Gemini Notebook, Teachable Machine, Amazon CloudWatch, n8n, KNIME, Zapier, and Trello. Code-based tools: Python, Google Colab, VS Code, Jupyter, OWASP Threat Dragon, Hugging Face, Databricks, DSPy, Garak, LangChain, LlamaIndex, and NVIDIA. The program also explores security frameworks and techniques, including MITRE ATLAS, OWASP Top 10 for LLMs, threat modeling, prompt injection assessments, and runtime guardrails. Note: This is an indicative list, and relevant tools are reviewed for inclusion periodically. Access to certain tools may require additional subscriptions or fees.

Eligibility & Enrollment

Do I need coding skills to learn AI security?

The program offers both code-based and no-code/low-code learning formats, catering to technology professionals as well as strategy and leadership professionals. The code-based format covers tools and technologies such as Python, Jupyter, LangChain, and Hugging Face, while the no-code/low-code format includes visual workflow tools such as n8n, KNIME, and Zapier. The preparatory module covers Python programming fundamentals, foundations of AI, and foundations of cybersecurity. The program does not specify prior coding experience as an eligibility requirement.

Can non-technical managers learn AI security without coding?

Yes. The program is designed for senior leaders, CXOs, and governance professionals who want to govern AI risks, ensure compliance, and evaluate secure AI architectures without writing code.

How do I enroll in the AI security program?

The enrollment involves a three-step process: complete an online application, have the application reviewed by a Great Learning panel, and receive an offer letter if selected. Contact Great Learning for current application requirements and available enrollment dates.

Fee & Payment

What is the fee for the Johns Hopkins University Certificate Program in AI Security?

The program fee is USD 1,800.

Are payment plans or financial assistance available for the AI security program?

Please contact a Program Advisor for information about payment plans, promotional offers, and financial assistance eligibility. Availability and applicable terms should be confirmed directly with the advisor.

Career-related Queries

What career opportunities can AI Cybersecurity skills support?

AI cybersecurity skills can support career development in areas such as AI security engineering, AI security assessment, application security, AI red teaming, AI risk management, and AI governance. Relevant opportunities depend on your existing experience, technical skills, and individual employer requirements.

Is AI security training relevant for professionals working with generative AI and AI agents?

Yes. Generative AI applications and AI agents can introduce risks through prompts, connected tools, external data sources, and autonomous actions. AI security training helps professionals identify these vulnerabilities, test defenses, establish access controls, and implement safeguards for LLM, RAG, and agentic systems.

How can AI security training help technology leaders manage AI risk?

AI security training can help technology leaders evaluate AI use cases, understand system-level vulnerabilities, assess third-party AI risks, establish human oversight, and align security practices with regulatory frameworks. The JHU AI Security program includes strategic risk assessment and no-code/low-code activities for leaders who oversee AI adoption.

General Queries

How is AI security different from traditional Cybersecurity?

Traditional cybersecurity focuses on protecting networks, applications, infrastructure, and data. AI security also addresses AI-specific risks, including jailbreaks, adversarial inputs, model manipulation, retrieval-augmented generation (RAG) vulnerabilities, and misuse of autonomous AI agents.

What is the difference between AI security and LLM security?

AI security covers the broader protection of AI systems, including models, data pipelines, infrastructure, applications, and autonomous workflows. LLM security focuses on risks specific to large language models, such as prompt injection, jailbreaks, sensitive data exposure, model misuse, and unsafe outputs. LLM security is an important part of the wider AI cybersecurity landscape.

Role-Based Queries

How can AI and data professionals learn to secure the AI systems they build?

AI and data professionals can develop AI security skills by learning threat modeling, secure LLM and RAG architecture, data integrity controls, model alignment, and runtime defense. The Johns Hopkins University program helps AI practitioners extend their expertise from building and deploying AI systems to identifying and mitigating vulnerabilities across the AI lifecycle.

What should Cybersecurity professionals learn to protect LLMs and AI agents?

Cybersecurity professionals should understand prompt injection, jailbreaks, data poisoning, model extraction, AI supply chain risks, and agentic tool abuse. They should also learn secure deployment, adversarial testing, runtime guardrails, and AI incident response. These topics are covered in the Johns Hopkins University Certificate Program in AI Security.

How can CTOs and CISOs evaluate AI security risks without coding?

CTOs and CISOs can evaluate AI security risks by assessing business value against security and privacy risks, reviewing AI architectures and third-party dependencies, establishing governance controls, and defining human oversight. The program's no-code/low-code activities are designed to help technology leaders make informed decisions about secure AI adoption.

How can AI leadership and strategy professionals ensure secure AI adoption?

AI leadership and strategy professionals can integrate security into AI planning by identifying risks early, assessing model and vendor dependencies, establishing responsible AI requirements, and defining appropriate oversight for automated systems. The program addresses these needs through strategic risk assessment, AI governance, secure architecture concepts, and practical no-code/low-code learning activities.

How can governance, risk, and compliance professionals prepare for AI regulations?

Governance, risk, and compliance professionals can develop relevant skills in AI risk assessment, responsible AI, supply chain governance, monitoring, and regulatory compliance. The program covers AI governance, global regulatory requirements, compliance monitoring, third-party risk management, and human oversight to help professionals understand and address AI-related governance responsibilities.

Delivered in Collaboration with:

Johns Hopkins University is collaborating with online education provider Great Learning to offer the Certificate Program in AI Security. Great Learning is a professional learning company with a global footprint in 170+ countries. Its mission is to make professionals around the globe proficient and future-ready. This program leverages JHU's leadership in innovation, science, engineering, and technical disciplines developed over years of research, teaching, and practice. Great Learning manages the enrollments and provides industry experts, student counselors, course support and guidance to ensure students get hands-on training and live personalized mentorship on the application of concepts taught by the JHU faculty.

Got more questions? Talk to us

Connect with our advisors and get your queries resolved

Speak with our expert +14109365774 or email to jhu-aisecurity@mygreatlearning.com

career guidance