CY0-001 study plan: pass CompTIA SecAI+ in 4 weeks

This plan assumes you already work in security and can give it 6–8 hours a week. If AI is entirely new to you, add a week at the front for concepts. If security is new to you, this is the wrong exam to start with — see CY0-001 vs Security+.

The plan follows the domain weights rather than the order they appear in the outline, because one domain is worth 40% and the others are not.

Week 1: AI concepts for security people (17%)

Get fluent in the vocabulary before trying to secure anything.

• Understand the lifecycle: training data, training, a model, deployment, inference. Know which stage an attack targets.
• Learn the failure modes that are specific to AI: hallucination, prompt injection, data poisoning, model extraction, evasion.
• Learn why generative systems differ from traditional software — non-determinism, natural-language input as an attack surface, and outputs that cannot be fully predicted.
• Get comfortable with the distinction the whole exam rests on: securing AI versus using AI to secure.

Weeks 2 and 3: Securing AI systems (40%)

Two full weeks, because this is two fifths of the exam.

Week 2 — data and models:

• Protecting training data: provenance, integrity, poisoning, and why data quality is a security concern.
• Protecting the model itself: theft, extraction, inversion, unauthorised fine-tuning.
• Protecting outputs: filtering, leakage of training data or secrets, and guarding against harmful content.

Week 3 — the environment:

• Securing deployment environments in cloud and on-premises.
• Identity and access for AI workloads: service identities over static keys, least privilege on model endpoints.
• Network exposure of inference endpoints, and rate limiting as a security control rather than a cost control.
• Logging and monitoring that is actually useful for AI: prompts, outputs, anomalies.
• Supply chain: pretrained models, third-party datasets, and the libraries around them.

Week 4: AI-assisted security (24%) and GRC (19%)

Two smaller domains together, then revision.

AI-assisted security:

• Where AI genuinely helps: detection at volume, triage, correlation, summarising incidents.
• Where it does not: anything requiring certainty, and anything where a false negative is unacceptable without review.
• The risks of defensive AI — over-trust, alert fatigue shifted rather than solved, and adversaries who know you use it.

Governance, risk and compliance:

• How AI risk is assessed and who owns it.
• The shape of the major frameworks and what they ask for: inventories, impact assessments, human oversight, documentation.
• Ethics and compliance obligations as operational requirements rather than abstractions.

Where the hours go

The last week

• Take a full practice exam under real conditions: 60 minutes, no notes, no pausing. The time pressure is part of what you are training for.
• Sort every mistake by domain. If the securing AI systems bar is weak, that alone can fail you.
• Practise the performance-based items specifically. Being slow on those is a more common cause of failure than not knowing the material.
• Take a second practice exam. Book when you are comfortably above 80% and finishing inside the hour.

If time runs short, protect weeks 2 and 3. Cutting a week from securing AI systems to revise governance is trading a 40% domain for a 19% one.