CY0-001 cheat sheet

Updated September 20, 2026

Last-minute reference for CY0-001. Securing AI systems is 40% of the exam, and the right answer removes an attack’s mechanism rather than merely detecting it.

Exam facts

Duration60 minutes
QuestionsMax 60
Passing score600 / 900
LaunchedFebruary 2026
PriceNot published by CompTIA; varies by region

Domain weights

DomainWeight
Securing AI systems40%
AI-assisted security24%
AI governance, risk, and compliance19%
Basic AI concepts related to cybersecurity17%

Attack names

AttackWhat it is
Prompt injectionInstructions smuggled into input; indirect arrives via retrieved content
JailbreakingPhrasing that circumvents guardrails
Data poisoningCorrupting training data to plant behaviour
Model extractionMany crafted queries to replicate the model
Membership inferenceDetermining whether a record was in training data
Adversarial evasionCrafted input that defeats detection at inference
Denial of walletExpensive requests flooding a paid endpoint

Attack to control

AttackControl that removes the mechanism
Indirect prompt injectionIsolate retrieved content from the instruction channel
Model extractionRate limiting, quotas, restrict output richness
Data poisoningProvenance checks and validation of contributed data
Excessive agent damageLeast-privilege tool scoping + confirmation on destructive actions
Secret disclosureKeep secrets outside the model’s reach entirely
Denial of walletRate limits, input size caps, per-tenant quotas
Weight theft in useConfidential computing

Governance essentials

  • Inventory first — owner, purpose, risk classification. Nothing else works without it.
  • Assess before deployment and on a schedule, not once.
  • The deploying organisation is accountable, regardless of who built the model.

Traps

  • Filtering keywords is not the fix for prompt injection; isolation is.
  • Instructing the model not to reveal secrets is not a control.
  • Auto-closing alerts hides false negatives — sample and review closures.
  • Output-only filtering leaves malicious input reaching the model, which may drive harmful tool use before any output exists.
  • This is not an entry point into security. It assumes the fundamentals.

Night-before checklist

  • Name every attack in the table above unprompted
  • For each, the control that removes its mechanism
  • Four domains and weights
  • Check ID and proctoring rules — see exam day

Take the 20-question practice test.