Format, cost, duration and level appear exactly as each provider states them. Courses are not scored, and we list a
missing value as not documented rather than guessing.
The Certified AI Security Professional (CAISP) is a course and certification from Practical DevSecOps. Its syllabus has seven chapters, including attacking LLMs, LLM Top 10 vulnerabilities, threat modeling AI systems and supply chain attacks in AI. Training is self-paced with browser labs and instructor support. The certification exam is online: candidates solve 5 challenges in 6 hours and then have 24 more hours to submit a report. The page states no formal difficulty level.
- Level
- Not documented in the sources we checked
Topics
- Understanding and attacking large language models
- LLM Top 10 vulnerabilities
- AI attacks and defenses using DevOps
- Threat modeling AI systems
- Supply chain attacks in AI
- Emerging threats, governance and compliance in AI
Best for: AppSec, DevSecOps and MLOps engineers who want a self-paced course with a practical certification exam on AI and LLM security
TAISE (Trusted AI Safety Expert) is a certificate from the Cloud Security Alliance, developed with Northeastern University. It has ten modules that range from AI basics and generative AI architecture to threat taxonomy, governance and compliance, cloud and AI security, and data security and privacy. CSA sells it only as a training plus exam bundle for $795, with two exam attempts. The exam has 60 multiple-choice questions in 120 minutes, and 80% is needed to pass.
- Duration
- Not documented in the sources we checked
- Level
- Not documented in the sources we checked
Topics
- AI model lifecycle and threat taxonomy
- Model governance, risk management and compliance
- Introduction to AI security and safety
- Cloud and AI security
- Data security and privacy in AI systems
- Ethics, transparency and explainability in AI
Best for: Professionals responsible for AI adoption, governance, compliance, risk or technical security who want a vendor-neutral certificate that bundles training and the exam
SecAI+ is a CompTIA certification exam (series code CY0-001) that launched on February 17, 2026. CompTIA lists four domains: basic AI concepts related to cybersecurity at 17%, securing AI systems at 40%, AI-assisted security at 24%, and AI governance, risk and compliance at 19%. The exam has up to 60 questions in 60 minutes, and the passing score is 600 on a scale of 100 to 900. CompTIA estimates retirement about three years after launch. The page shows no price.
- Duration
- Not documented in the sources we checked
- Level
- Not documented in the sources we checked
Topics
- Basic AI concepts related to cybersecurity (17% of the exam)
- Securing AI systems (40% of the exam)
- AI-assisted security (24% of the exam)
- AI governance, risk and compliance (19% of the exam)
Best for: IT and cybersecurity staff who hold or are working toward CompTIA credentials and want an exam that weights securing AI systems most heavily
This IBM specialization on Coursera teaches generative AI skills for cybersecurity careers. It has three courses: generative AI introduction and applications, prompt engineering basics, and generative AI for a cybersecurity career. Coursera marks it Intermediate, at 12 weeks and 5 hours a week, and says no prior experience is necessary. The page lists skills such as incident response, threat detection and SIEM, and it offers a shareable certificate. The page states no price.
- Exam
- Not documented in the sources we checked
Topics
- Generative AI introduction and applications
- Prompt engineering basics
- Generative AI for a cybersecurity career
- Listed skills include incident response, threat detection, threat modeling and SIEM
Best for: Security practitioners and newcomers who want a structured introduction to generative AI and prompt engineering applied to security tasks
Red Teaming LLM Applications is a short DeepLearning.AI course made with Giskard and taught by two Giskard staff. It has seven video lessons, one reading and a graded quiz, 1h 39m in total, and is marked Beginner. Lessons cover LLM vulnerabilities, red teaming chatbots, red teaming at scale, using LLMs to red team LLMs with Giskard's open-source library, and a full assessment. Enrollment is free, and a graded assignment needs PRO membership.
Topics
- Overview of LLM vulnerabilities
- Red teaming LLMs, including prompt injection attacks on chatbots
- Red teaming at scale
- Red teaming LLMs with LLMs using Giskard's open-source library
- A full red teaming assessment
Best for: Developers and security staff who want a short beginner-level introduction to red teaming LLM applications and know basic Python
AI Red Teamer is a Hack The Box Academy job role path of 12 modules, built in collaboration with Google. The modules run from AI fundamentals through prompt injection, LLM output attacks, AI data attacks, attacks on AI applications and MCP, evasion attacks, AI privacy and AI defense. HTB marks the path Hard. The path page states no duration, price or prerequisites. HTB's certifications page lists a separate COAE certification for this path.
- Duration
- Not documented in the sources we checked
- Prerequisites
- Not documented in the sources we checked
Topics
- AI fundamentals (Module 1) and building AI models for infosec tasks such as spam, network anomaly and malware classification (Module 2)
- Introduction to red teaming AI (Module 3) and prompt injection attacks, with jailbreak research (Module 4)
- LLM output attacks, including improper handling of LLM outputs and LLM abuse
- AI data attacks on training data and data pipelines
- Attacking AI applications and systems, including the Model Context Protocol
- AI evasion attacks, AI privacy attacks and AI defense
Best for: Offensive security practitioners who want a structured path from AI fundamentals to attacking LLM applications and machine learning models
AIGP (Artificial Intelligence Governance Professional) is an IAPP certification focused on AI governance. IAPP describes it as covering AI systems and responsible AI principles, how laws apply to AI, and the AI life cycle and risk management. The exam has 100 questions and is taken through Pearson VUE, at a test center or remotely. IAPP lists US$649 for members and US$799 for non-members, and its store page states no prerequisites.
- Duration
- Not documented in the sources we checked
- Level
- Not documented in the sources we checked
- Prerequisites
- Not documented in the sources we checked
Topics
- Foundational knowledge of AI systems and responsible AI principles
- How current and emerging laws apply to AI
- The AI life cycle and risk management
- Awareness of debated AI governance issues
Best for: People who work on AI governance and want a credential covering responsible AI principles, AI laws and risk management across the AI life cycle
AAISM (Advanced in AI Security Management) is an ISACA certification aimed at people in security or advisory roles who work with AI systems. ISACA's page lists three practice areas: AI governance and program management, AI technologies and controls, and AI risk management. Candidates must already hold an active CISM or CISSP, pass a computer-based exam at a PSI center or remotely, and pay a US$50 application fee. The page does not state the question count or exam length.
- Duration
- Not documented in the sources we checked
- Level
- Not documented in the sources we checked
Topics
- AI governance and program management
- AI technologies and controls
- AI risk management
Best for: Holders of CISM or CISSP who want a credential covering AI governance, AI risk management and AI controls
The AI Red Teaming, Prompt Hacking and AI Security Masterclass is an on-demand course from Learn Prompting that comes with an AIRTP+ certification exam. Learn Prompting lists about 25 hours of material in four modules: AI red teaming basics, prompt hacking and vulnerabilities, advanced red teaming and defense, and automation with tools such as PyRIT. The $1,199 price includes an exam voucher. The exam is a 24-hour practical assessment, and passing it earns the certification.
- Level
- Not documented in the sources we checked
Topics
- Introduction to AI red teaming, classical security principles and the OWASP Top 10 for LLMs
- Prompt hacking techniques and AI vulnerabilities
- Advanced red teaming and AI defense strategies
- Automated red teaming tools such as PyRIT, and advanced prompt injection
Best for: People who want a self-paced prompt hacking and AI red teaming course that ends with a practical certification exam
AI security fundamentals is a free Microsoft Learn path of three modules. They cover how AI security differs from traditional security and AI-specific attacks such as jailbreaking and prompt injection, security controls such as content filtering, metaprompts and grounding, and an introduction to AI security testing and red teaming. Microsoft marks the path Beginner and lists basic AI and security concepts as the prerequisite. The page text does not state a total duration.
- Duration
- Not documented in the sources we checked
- Exam
- Not documented in the sources we checked
Topics
- AI-specific attacks: jailbreaking, prompt injection, model manipulation, data exfiltration and overreliance
- AI security controls: supply chain security, content filtering, metaprompts, grounding and monitoring
- AI red teaming and the three categories of AI security testing
Best for: Developers, administrators and security engineers who want a free introduction to AI security threats, controls and red teaming
PortSwigger's Web Security Academy has a free section on web LLM attacks. It explains how attackers abuse language models that are wired into web applications, covering prompt injection, calls to LLM APIs and functions, insecure output handling, indirect prompt injection and training data poisoning, and it covers defenses. The Academy offers interactive labs for practice, and an account is used to track progress. The provider does not state a duration, level or exam for this section.
- Duration
- Not documented in the sources we checked
- Level
- Not documented in the sources we checked
- Exam
- Not documented in the sources we checked
- Prerequisites
- Not documented in the sources we checked
Topics
- LLM attacks and prompt injection
- Exploiting LLM APIs, functions and plugins
- Insecure output handling
- Indirect prompt injection
- Training data poisoning
- Defending against LLM attacks
Best for: Web developers and penetration testers who want free practice labs for attacks on LLM-backed web apps
SEC545 is an advanced SANS course on securing generative AI and LLM applications. SANS lists five syllabus sections: GenAI and LLM security risks, securing GenAI applications, agentic AI security, MLSecOps and the application lifecycle, and AI for security. It runs 5 days instructor-led or 30 hours self-paced, and it links to the GIAC AI Platform Security (GAIPS) certification. SANS lists US$8,260 for the OnDemand format.
Topics
- GenAI, large language models and security risks
- Securing GenAI applications
- Agentic AI security
- MLSecOps and securing the GenAI application lifecycle
- AI for security
Best for: Security practitioners who want a structured advanced course on securing LLM applications, agents and MLOps pipelines, with a linked GIAC certification
Choose by role and by what you need to show. Labs and games build skill through practice, courses add structure, and certifications give a credential others can check.
Setup
- Start from the role. Security engineers and red teamers usually want labs they can work through. Security managers and governance leads often want a course or certification built around frameworks. ML engineers gain most from material on the model supply chain and on evaluation.
- Decide between practice and credential. A lab builds a skill; a certification shows you passed a published syllabus.
- Check level and prerequisites. If you are new to LLM security, start with free material.
Integration
Limits
Cost and duration appear exactly as each provider states them, with currency where the page gives one. Prices change, so confirm on the provider's page before you buy. We do not rank or score courses, we do not take them, and a listing is not an endorsement.
