Skip to content

Get the app

Study guide · Domain 2.0

Threats, Vulnerabilities, and Attacks

24% of the exam · 6 exam objectives

Domain 2 is the largest share of the exam (24%). It asks you to describe who attacks and why, how attacks get in, what makes a system vulnerable, how to read the signs of an attack in logs and behaviour, and which threats come with the use of AI. Most questions give a short situation and ask you to name the threat, vector, vulnerability or attack it shows.

A study feature written by ExamDragon, not part of the real exam. Each heading is one of this domain's 6 exam objectives, as CompTIA states it in the SY0-801 objectives.

2.1Explain characteristics of threats and vulnerabilities.

  • Risk from a threat combines likelihood (how probable it is) with impact (how much harm it causes).
  • Threat feeds deliver indicators such as malicious IP addresses and domains in a form tools can ingest; intelligence sources range from open-source to proprietary, shared and internal data.
  • The intelligence life cycle runs from requirements to collection, processing, analysis, dissemination and feedback.
  • Common Vulnerabilities and Exposures (CVE) gives each disclosed vulnerability a unique identifier; the Common Vulnerability Scoring System (CVSS) scores its severity from 0.0 to 10.0.
  • CVSS base metrics include the attack vector, privileges required and user interaction, and the impact on confidentiality, integrity and availability.
  • Prioritization adds context to the score: exposure of the asset, active exploitation and business impact can move a lower score ahead of a higher one.
  1. The situationA scan returns a 9.1 finding on an isolated test server and a 7.2 finding on the internet-facing payment portal that a threat feed says is being exploited.
  2. Step 1Read the base scores, but treat them as a starting point.
  3. Step 2Check exposure: the portal is reachable from the internet, the test server is not.
  4. Step 3Check likelihood: the feed reports active exploitation of the portal flaw.
  5. Step 4Fix the portal first, then schedule the test server.
  6. Where questions turnTreating the CVSS base score as the whole of prioritization, or mixing up CVE (the name) with CVSS (the score).
  7. The takeawayBase score plus exposure, exploitation and impact sets the order of work, not the base score alone.
Practise 2.1

2.2Describe common threat actors and motivations.

  • Actors listed: crime syndicate or organized crime, terrorist, unskilled attacker, hacktivist, insider, accidental or unintentional, competitor and state-sponsored.
  • Motivations include financial, influence, intellectual property, notoriety, espionage, fear and chaos, extortion, general curiosity, revenge, ideological, political and ethical.
  • Actors are described by three attributes: internal or external, resources and funding, and level of sophistication or capability.
  • Crime syndicates want money; hacktivists want a cause seen; state-sponsored actors pursue espionage or strategic aims with large resources.
  • Insiders hold trusted access; an accidental or unintentional threat causes harm by mistake, with no intent.
  • Unskilled attackers use tools written by others and are often driven by notoriety or curiosity.
  1. The situationIntruders kept hidden access to a research lab for 18 months using tools never seen before and copied vaccine data.
  2. Step 1Attribute: they came from outside, so external.
  3. Step 2Capability: custom tools and long stealth mean high sophistication and large resources.
  4. Step 3Motivation: copying research data is espionage.
  5. Step 4Conclusion: a state-sponsored actor is the best fit.
  6. Where questions turnNaming the actor from the technique alone. Data theft can be a competitor, an insider or a state; the motivation and attributes decide.
  7. The takeawayCombine motivation with the three attributes to name the actor.
Practise 2.2

2.3Describe threat vectors and sources.

  • Message-based vectors are email, Short Message Service (SMS), Rich Communication Services (RCS), instant messaging and collaboration tools; image-based ones are quick response (QR) codes, embedded content and CAPTCHA.
  • Attachment-based vectors include embedded macros, Rich Text Format (RTF) documents and Portable Document Format (PDF) files; browser-based ones include extensions, JavaScript, password managers, cookies and session tokens.
  • Network-based vectors are infrastructure devices, virtualized devices and session keys; remote access covers remote desktop, Virtual Network Computing (VNC) and virtual private networks (VPN).
  • Endpoint-based vectors include mobile devices, workstations, servers, tablets, trusted devices, built-in tools and living-off-the-land tools; an advanced persistent threat (APT) often relies on these to stay hidden.
  • Supply chain-based vectors run through third-party, managed service, logistic and software as a service (SaaS) providers; external media covers malicious Universal Serial Bus (USB) devices.
  • Human-based vectors are impersonation, contractors, visitors, biometrics and watering holes; physical-based ones are lock and key, access vestibules and access passes.
  • Internet of Things (IoT) devices such as cameras, sensors and printers, operational technology (OT), and signal-based links (Bluetooth, radio frequency, near-field communication) are also vectors.
  1. The situationStaff of a law firm often visit a legal news site. Attackers compromise the site, and visitors from the firm receive malware.
  2. Step 1Ask where the attack started: on a site the victims already trust.
  3. Step 2Note the attacker did not contact the victims directly, so this is not message-based.
  4. Step 3Name it: a watering hole, a human-based vector.
  5. Where questions turnConfusing the vector (how the attack arrives) with the attack itself, for example naming a QR code sticker as a malware type instead of an image-based vector.
  6. The takeawayName the vector by the path the attack took to reach the victim.
Practise 2.3

2.4Explain types of vulnerabilities and attack surfaces.

  • Unsupported products get no more vendor fixes; unpatched systems have fixes available that are not applied; obsolete and unmanaged systems lack upkeep and oversight.
  • Application flaws include race conditions between time-of-check (TOC) and time-of-use (TOU) and malicious updates delivered through trusted channels.
  • Code flaws include hardcoded secrets and unsafe exception handling that exposes internal details.
  • A zero-day is a flaw unknown to the vendor with no fix yet; cryptographic vulnerabilities come from weak or misused cryptography.
  • Unmanaged or stale credentials, rogue devices and shadow IT are weaknesses of oversight rather than code.
  • Other attack surfaces: ports and services, operating systems, virtualization, wireless and low-powered communications, large language models (LLM), identity providers, mobile devices, misconfiguration, public repositories and public object storage.
  1. The situationAn error page on a staff portal shows a stack trace that includes a database password.
  2. Step 1Ask what reveals the secret: the way the error is shown to the user.
  3. Step 2Rule out a hardcoded secret as the main finding, because the leak happens through the error page.
  4. Step 3Name it: unsafe exception handling, fixed by generic messages and internal logging.
  5. Where questions turnTreating unsupported and unpatched as the same thing. Unpatched means a fix exists; unsupported means none will come.
  6. The takeawayFind the step that makes the weakness exploitable and name that.
Practise 2.4

2.5Given a scenario, analyze indicators of malicious activity.

  • Malware listed: ransomware, Trojan, worm, spyware, adware, virus, rootkit, keylogger, logic bomb and fileless malware.
  • Physical attacks are tailgating, shoulder surfing, skimming and forced entry; network attacks include distributed denial of service (DDoS), protocol downgrade, rogue devices, sniffing, spoofing, on-path, Domain Name System (DNS) attacks and cache poisoning.
  • Social engineering attacks are smishing, vishing, phishing, whaling, spear phishing, quishing, impersonation and deepfake.
  • Indicators of compromise include hashes, IP addresses, domains, malicious processes, file system artifacts, timestamps, log manipulation, excessive resource consumption, plaintext strings, account lockout, impossible travel and concurrent sessions.
  • Application attacks are injection, buffer overflow, replay, privilege escalation, forgery and directory traversal.
  • Credential attacks are password spraying, brute force, user enumeration, replay and multifactor authentication (MFA) bypass.
  1. The situationLogs show 20 failed sign-ins on one account in a minute, then an account lockout.
  2. Step 1Note the pattern: many passwords against one account.
  3. Step 2Contrast with spraying, which tries one password across many accounts.
  4. Step 3Name it: brute force, with account lockout as the indicator.
  5. Where questions turnMixing up password spraying and brute force, or the social engineering channels: smishing is text, vishing is voice, quishing is a QR code.
  6. The takeawayThe shape of the activity in the logs names the attack.
Practise 2.5

2.6Summarize threats and vulnerabilities associated with artificial intelligence (AI) usage.

  • Prompt injection plants instructions in input the model processes; jailbreaking persuades a model to bypass its own safety rules.
  • Poisoning corrupts training data; evasion crafts inputs that fool a trained model; model manipulation alters the model itself.
  • Data loss happens when sensitive data is entered into AI tools outside the organization's control; privacy failures include a model revealing personal data.
  • Bias is unfair skew in results; explainability is whether anyone can show why a model decided as it did; ethical considerations cover fair and responsible use.
  • Hallucinations are confident but false outputs, such as invented facts or software packages.
  • Session hijacking can take over a user's AI sessions, and code execution risk arises when model output is run as commands.
  1. The situationA support chatbot reads customer emails. One email contains hidden text telling it to reveal the discount codes it was told to keep private.
  2. Step 1Ask where the hostile instruction came from: content the model reads, not the model's training.
  3. Step 2Rule out poisoning, which happens before deployment.
  4. Step 3Name it: prompt injection, reduced by treating content as data and limiting what the bot can reveal or do.
  5. Where questions turnTreating prompt injection, jailbreaking and poisoning as the same thing. They differ in who supplies the input and at which stage.
  6. The takeawayAsk when the attack acts: on training data (poisoning), on live input (prompt injection, jailbreaking, evasion) or on the model itself (model manipulation).
Practise 2.6

Key terms

APT
Advanced Persistent Threat
CVE
Common Vulnerabilities and Exposures
CVSS
Common Vulnerability Scoring System
DDoS
Distributed Denial of Service
LLM
Large Language Model
MFA
Multifactor Authentication
NFC
Near-field Communication
RCS
Rich Communication Services
TOC
Time-of-check
TOU
Time-of-use
After this guide: Logs show one password tried once against hundreds of accounts. Is this brute force or password spraying, and what single control would most reduce the chance it succeeds?
Show the answer

Password spraying, because it is one password across many accounts. Multifactor authentication most reduces its success, since a guessed password alone is no longer enough.

Practise Threats, Vulnerabilities, and Attacks

Built to CompTIA's Security+ SY0-801 V8 exam objectives. The explanations, worked examples and traps are ExamDragon's own. Every source this path is built on.