---
title: "Security Fundamentals: CIA, Threats and Risk"
domain: cybersecurity
level: beginner
duration: "40 min"
objectives:
  - "Explain the CIA triad (confidentiality, integrity, availability) and give an example of a failure of each."
  - "Distinguish assets, threats, vulnerabilities and risk, and identify common threat actors."
  - "Estimate a risk as likelihood × impact and rank risks in a simple risk register."
  - "Classify security controls by purpose (preventive, detective, corrective) and by type (technical, administrative, physical)."
  - "Explain defence in depth and least privilege, and apply them to a small scenario."
---

# Security Fundamentals: CIA, Threats and Risk

## Why this matters
A small online shop gets an e-mail that looks like it comes from its payment provider: "Your account is locked, log in here." The owner's assistant clicks, types the shop's admin password into a fake page, and goes back to work. Two days later, customer orders are being changed, and a list of names and addresses is for sale. Nothing about this attack was clever. It worked because nobody had asked three simple questions: what do we need to protect, what could go wrong, and what are we doing about it?

Security is not the art of stopping every possible attack. No organisation has the money or the time for that. It is the habit of protecting **what matters most**, against the threats that are **most likely**, with controls you can **afford**. This lesson gives you the vocabulary and the way of thinking that every other security topic builds on: the CIA triad, the difference between a threat, a vulnerability and a risk, how to put a number on risk, and how to choose and layer controls.

> [!NOTE]
> This course teaches defence. Attacks are described so that you can recognise and prevent them, never so that you can carry them out. Only ever test systems that you own or are explicitly authorised to test.

## Learning objectives
- Explain the CIA triad (confidentiality, integrity, availability) and give an example of a failure of each.
- Distinguish assets, threats, vulnerabilities and risk, and identify common threat actors.
- Estimate a risk as likelihood × impact and rank risks in a simple risk register.
- Classify security controls by purpose (preventive, detective, corrective) and by type (technical, administrative, physical).
- Explain defence in depth and least privilege, and apply them to a small scenario.

## Key concepts

### The CIA triad
Imagine the paper files of a doctor's surgery. They must be seen only by the right people (the cabinet is locked), they must be correct (nobody scribbles over a diagnosis), and they must be there when a patient needs them (the cabinet isn't lost in a fire). Those three needs are the **CIA triad**, the classic model of what security protects:

- **Confidentiality** means information is available only to the people and systems allowed to see it. A failure is a **data breach**: a customer list leaked, a password database stolen, an e-mail sent to the wrong person.
- **Integrity** means information and systems are accurate, complete and changed only by those allowed to change them. A failure is tampering: a price changed in a database, a bank transfer altered in transit, malware modifying a file.
- **Availability** means systems and data are usable when they are needed. A failure is an outage: a shop that is down on a sale day, files locked by ransomware, a server taken offline by a flood of traffic (a **distributed denial of service**, or DDoS, attack).

![Three cards for the CIA triad. Confidentiality: only the right people can see the information; failure, a customer list is leaked; control, encryption and access control. Integrity: data is accurate and not altered without permission; failure, a price is changed in the database; controls, hashes, change approval and backups. Availability: systems and data are usable when needed; failure, the shop is down on a sale day; controls, redundancy, backups and DDoS protection. A panel below adds authenticity (trusting who or what something really is) and non-repudiation (not being able to deny an action you took, thanks to logs or signatures).](../../../assets/diagrams/secfund-cia-triad.svg)

*Figure 1. Each property has its own kind of failure and its own kind of control. One incident can break several at once.*

The three properties often pull against each other. Locking everything down tightly helps confidentiality but can hurt availability: if only one person holds the key and she is on holiday, nobody can work. Part of a security professional's job is finding the right balance for each asset. For a hospital's patient record, availability and integrity may matter even more than confidentiality; for a secret product design, confidentiality comes first.

Two related properties are often added to the triad:

- **Authenticity** means you can trust who or what something really is: that a message really comes from the sender it claims, or that a file is the genuine one.
- **Non-repudiation** means a person cannot credibly deny an action they took, because there is reliable evidence such as logs or a digital signature.

Later modules (cryptography, authentication) show the tools that provide them.

> [!CHECK]
> An attacker changes the delivery address in a customer's order in the shop database, so the parcel goes to the wrong house. Which CIA property failed?
>
> - [ ] Confidentiality, because the attacker saw the address
> - [x] Integrity, because data was changed without permission
> - [ ] Availability, because the order was affected
> - [ ] None, because no data was stolen
>
> Integrity is about data being accurate and changed only by authorised people. Nothing was leaked and the system kept running; the data was altered.

### Assets, threats, vulnerabilities and risk
People often use these four words as if they meant the same thing. They don't, and mixing them up leads to muddled decisions.

- An **asset** is anything of value that you want to protect: data, a system, a service, a person's time, your reputation. Start here, because you cannot protect what you haven't listed.
- A **threat** is anything that could cause harm to an asset: a criminal, a power cut, a careless employee, a flood.
- A **vulnerability** is a weakness that a threat can use: unpatched software, a weak password, an unlocked server room, a missing backup.
- A **risk** is the chance that a threat uses a vulnerability to harm an asset, combined with how bad that harm would be.

An everyday example: your house is the asset, a burglar is the threat, an unlocked back door is the vulnerability, and the risk is how likely a burglary is, times how much it would cost you. Lock the door and the vulnerability is gone, so the risk falls. The burglar still exists, though. You usually can't remove threats; you reduce vulnerabilities and impact.

![Three cards in a row. A threat, something that could cause harm, for example a phishing email, exploits a vulnerability, a weakness the threat can use, for example no second login step (multi-factor authentication, MFA), which exposes an asset, something of value to protect, for example customer data and orders. Below, a Risk panel says risk is the chance that a threat uses a vulnerability to harm an asset and how bad that would be: risk equals likelihood times impact. A Controls panel says controls are safeguards that lower the likelihood or the impact.](../../../assets/diagrams/secfund-risk-relationship.svg)

*Figure 2. A threat only becomes a risk when it can reach an asset through a vulnerability. Controls break that chain.*

#### Threat actors
A **threat actor** is a person or group that can cause harm on purpose. Knowing who they are tells you what they want and how much effort they will spend.

| Actor | Typical motive | Typical resources | Example |
|---|---|---|---|
| **Criminals** | Money | Anything from cheap automated tools to organised gangs | Ransomware, stolen card data, fake invoices |
| **Insiders** | Grievance, money, or none (mistakes) | Legitimate access, which is what makes them dangerous | A leaver copying the customer list; an employee sending a file to the wrong person |
| **Hacktivists** | A political or social cause | Usually modest, often public attacks | Defacing a website, leaking documents, DDoS |
| **Nation states** | Espionage, disruption, strategic advantage | Very large: skilled teams, long campaigns | Long-term access to a government or energy network |

Not every threat is a person. Accidents and natural events count too, and in practice they cause a large share of incidents. Also notice that most small organisations are hit by the first row, by opportunistic criminals who attack whoever is easy, not by someone who chose them.

#### Common threats
You will meet the same few threats again and again:

- **Phishing:** a message that tricks you into giving away a password, opening a harmful file or paying a fake invoice. It works on people, not on machines.
- **Ransomware:** malware that encrypts your files and demands payment for the key. It attacks availability, and many gangs also steal the data first, which attacks confidentiality.
- **Misconfiguration:** a system set up in an unsafe way, for example a cloud storage bucket left public or a default password never changed. No attacker skill is needed; they only have to look.
- **Lost or stolen devices:** a laptop left on a train. Whether it is a crisis or a shrug depends on whether its disk was encrypted.

> [!CHECK]
> A shop stores its orders in cloud storage that was accidentally set to "public". Which statement is correct?
>
> - [ ] The threat is the public setting
> - [ ] The asset is the misconfiguration
> - [x] The public setting is the vulnerability, and the order data is the asset
> - [ ] The risk is the order data
>
> The setting is the weakness (vulnerability). The data is what has value (asset). The threat would be anyone who finds and downloads it, and the risk combines how likely that is with how harmful it would be.

### Measuring risk
You can't fix everything at once, so you need a way to decide what comes first. A common, simple method gives each risk two scores from 1 to 5:

- **Likelihood:** how probable is it that this happens in the next year or so? (1 = rare, 5 = almost certain)
- **Impact:** how bad would it be if it did? Think of money, legal trouble, lost customers and lost time. (1 = minor, 5 = severe)

The **risk score** is **likelihood × impact**, from 1 to 25. A **risk matrix** turns the score into a band that everyone understands. In this course: 1–7 is **low**, 8–14 is **medium** and 15–25 is **high**. Organisations choose their own bands; what matters is that they apply them consistently.

![A five by five risk matrix. Likelihood runs from 1 rare to 5 almost certain on the vertical axis and impact from 1 minor to 5 severe on the horizontal axis. Each cell shows likelihood times impact. Scores 1 to 7 are low and green, 8 to 14 are medium and amber, 15 to 25 are high and red. The cell for likelihood 4 and impact 4, score 16, is circled as a worked example: phishing steals the admin password, a high risk.](../../../assets/diagrams/secfund-risk-matrix.svg)

*Figure 3. The same score can come from different pairs: 3 × 4 and 4 × 3 are both 12, but the first is a possible major harm and the second a likely serious one. Look at both numbers, not just the product.*

A **risk register** is simply a table that lists each risk with its likelihood, impact, score and, later, who owns it and what is being done. Sorted by score, it tells you where to spend your first euro and your first hour. In the lab you build one.

| Risk | Likelihood | Impact | Score | Band |
|---|---|---|---|---|
| Phishing steals the admin password | 4 | 4 | 16 | High |
| Hosting outage for a day | 3 | 3 | 9 | Medium |
| Old freelancer account still active | 2 | 3 | 6 | Low |

The numbers are judgements, not measurements. Two people may pick different scores, and that is fine: the value is in the discussion and in ranking, not in the exact figure. This is a **qualitative** method. Larger organisations also use **quantitative** methods that estimate money, for example the expected yearly loss.

#### Risk treatment
Once you know a risk, you have four basic choices:

| Treatment | What you do | Example |
|---|---|---|
| **Mitigate** (reduce) | Add controls to lower likelihood or impact | Turn on multi-factor authentication (MFA) for the admin account |
| **Transfer** (share) | Move part of the financial impact to someone else | Buy cyber insurance; use a payment provider so you never store card numbers |
| **Avoid** | Stop the activity that creates the risk | Stop keeping customers' dates of birth, because you don't need them |
| **Accept** | Knowingly live with it, usually because it is low or the fix costs more than the harm | Accept a low risk of a one-hour website outage |

Transferring a risk shifts the *cost*, not the *responsibility*: if your provider leaks your customers' data, you still have to tell your customers. Acceptance should be a written decision by someone with the authority to make it, not a result of never looking.

What is left after treatment is the **residual risk**. It is never zero.

> [!CHECK]
> A shop stops collecting customers' dates of birth because it never uses them. Which risk treatment is this?
>
> - [ ] Mitigate
> - [ ] Transfer
> - [x] Avoid
> - [ ] Accept
>
> The shop removes the activity that creates the risk. Mitigating would keep the data and protect it better; transferring would move the cost to an insurer.

### Security controls
A **security control** (also called a safeguard or countermeasure) is anything that reduces a risk. Controls are classified in two ways that are worth keeping separate.

**By purpose: what the control does.**

| Purpose | Question it answers | Examples |
|---|---|---|
| **Preventive** | How do we stop it happening? | MFA, a firewall, locked doors, security training |
| **Detective** | How do we notice it happening or having happened? | Alerts on failed logins, log review, cameras, file-integrity checks |
| **Corrective** | How do we limit the damage and recover? | Restoring from backup, patching, an incident response plan, disabling a stolen account |

Two more purposes appear in some frameworks. **Deterrent** controls discourage people from trying (a warning sign, a visible camera). **Compensating** controls stand in when the ideal control isn't possible (extra monitoring on an old system that can't be patched).

**By type: what the control is made of.**

| Type | Also called | Examples |
|---|---|---|
| **Technical** | Logical | Encryption, firewalls, antivirus, access control lists |
| **Administrative** | Managerial, procedural | Policies, training, background checks, incident procedures |
| **Physical** | Operational/physical | Locks, badges, fences, CCTV, a safe |

The two views combine. A camera is *physical* and *detective*. Security awareness training is *administrative* and *preventive*. An encrypted backup is *technical* and *corrective*. A good set of controls has a mix of purposes: prevention alone fails the first time something slips through, and without detection you may not find out for months.

> [!CHECK]
> A server sends an alert to the administrator whenever someone logs in from a new country. How is this control best classified?
>
> - [x] Technical and detective
> - [ ] Administrative and preventive
> - [ ] Technical and corrective
> - [ ] Physical and detective
>
> It is software (technical) and it notices suspicious activity (detective). It doesn't stop the login and it doesn't repair anything.

### Defence in depth and least privilege
**Defence in depth** means using several independent layers of controls, so that if one fails, another still stands. Think of a castle: a moat, a wall, a gate, guards and an inner keep. No single layer is perfect, and an attacker must beat all of them. In an organisation, the layers include policies and people, physical security, the network, devices, applications and the data itself.

![A table of six layers, from the outside in, that an attacker has to get past. People and policy: security training and least privilege, regular access reviews, administrative. Physical: locked door and badge reader, camera and visitor log, physical. Network: firewall and network segments, alerts on unusual traffic, technical. Device: disk encryption and updates, antivirus and device alerts, technical. Application: input checks and MFA login, alerts on failed logins, technical. Data: encryption and access by role, access logs reviewed, technical. A note says corrective controls, tested backups and an incident response plan, work across all layers.](../../../assets/diagrams/secfund-defence-in-depth.svg)

*Figure 4. Every layer has a preventive and a detective control. Corrective controls, such as tested backups, sit behind all of them.*

The layers should be **independent**: a firewall and a second firewall from the same vendor with the same flaw count as one layer. Layers should also cover different failure modes, such as people making mistakes, software having bugs and devices being lost.

**Least privilege** means every person, program and system gets only the access it needs to do its job, and no more, and only for as long as it is needed. A shop assistant who prepares parcels needs to read orders; she doesn't need to change prices or export the whole customer list. If her account is phished, the damage is limited by what that account can do. Least privilege limits both mistakes and attacks. It also reduces the damage an insider can do.

Putting it into practice means giving new accounts minimal access and adding more on request, reviewing access regularly, removing it when someone changes role or leaves, and using separate admin accounts only for admin tasks. A related idea is **separation of duties**: no single person can complete a critical action alone, such as one person raising a payment and another approving it.

> [!CHECK]
> A shop assistant only prepares parcels, but her account can also edit prices and export all customers. What principle is being broken?
>
> - [ ] Defence in depth
> - [x] Least privilege
> - [ ] Non-repudiation
> - [ ] Availability
>
> Her account has far more access than her job needs. If it is phished or misused, the damage is much bigger than it should be.

## How it works
Putting the concepts together gives a loop that organisations repeat all the time. It is a simplified version of the process in NIST SP 800-30:

1. **Identify assets.** List what you have and decide which CIA property matters most for each.
2. **Identify threats and vulnerabilities.** Ask what could go wrong and which weaknesses allow it.
3. **Assess risk.** Score likelihood and impact, and write the risks in a register, sorted by score.
4. **Treat risk.** For the top risks, choose to mitigate, transfer, avoid or accept. When you mitigate, pick a mix of preventive, detective and corrective controls, layered, with least privilege.
5. **Monitor and repeat.** Check that the controls work, record what changed, and update the register. New systems, new staff and new threats change the scores.

Larger organisations organise this work with a framework. The **NIST Cybersecurity Framework (CSF) 2.0**, published in 2024, groups security outcomes into six functions:

| Function | Meaning |
|---|---|
| **Govern** | Set the strategy, roles, policies and risk appetite |
| **Identify** | Know your assets, suppliers and risks |
| **Protect** | Put safeguards in place to reduce the likelihood and impact |
| **Detect** | Find attacks and incidents quickly |
| **Respond** | Act on a detected incident |
| **Recover** | Restore normal operation and learn from it |

The loop above covers Identify and Protect most directly. The later modules of this course go deeper into Detect, Respond and Recover.

> [!CHECK]
> In the risk process, what comes before choosing controls?
>
> - [ ] Buying security software
> - [x] Listing assets and ranking risks
> - [ ] Accepting every risk
> - [ ] Writing a policy for every threat
>
> Controls should follow the risks that matter. Without a list of assets and a ranked register, you may spend money on a minor problem and miss a major one.

## Worked example
A bakery, *Sunrise Bakery* (invented), sells bread online, with deliveries. It has a web shop, a customer list of about 800 names and addresses, and a laptop that the owner uses for orders and accounts.

**Assets and what matters most.**

| Asset | Most important property | Why |
|---|---|---|
| Customer list | Confidentiality | A leak harms customers and breaks data-protection rules |
| Web shop | Availability | No site means no sales during the morning rush |
| Order records | Integrity | Wrong addresses or prices mean losses and angry customers |

**Two risks scored.**

- *The owner's laptop is stolen from a café.* Threat: thief. Vulnerability: the disk isn't encrypted. Likelihood 3 (possible), impact 4 (it has the customer list and accounts). Score 3 × 4 = **12, medium**.
- *The shop's admin password is phished.* Threat: criminal. Vulnerability: no MFA. Likelihood 4, impact 4. Score **16, high**.

**Treatment for the high risk** (mitigate, in layers):

| Purpose | Control | Type |
|---|---|---|
| Preventive | Turn on MFA for the admin account | Technical |
| Preventive | A short training session on spotting phishing | Administrative |
| Detective | An e-mail alert on each new admin login | Technical |
| Corrective | A written plan: who resets passwords and checks orders, with a tested backup | Administrative and technical |

Least privilege applies too: only the owner holds the admin account, and the helper who packs orders gets a separate account that can see orders but not change prices. After the changes, the likelihood falls from 4 to 2, so the **residual risk** is 2 × 4 = 8, medium. It isn't zero, and the owner knowingly accepts that.

## Common pitfalls
- **Treating security as a product** — buying a tool doesn't make you secure. Controls need people and procedures, and they need to be checked.
- **Starting with controls instead of assets** — you end up protecting what is easy to protect, not what matters. List assets and risks first.
- **Focusing only on confidentiality** — a shop that is secret but offline or full of wrong data has failed too. Check all three CIA properties.
- **Relying on one layer** — a firewall alone, or training alone, fails the first time it is bypassed. Use layers.
- **Prevention without detection** — if you can't tell that something went wrong, an attacker can stay for months. Add detective and corrective controls.
- **Scores that never change** — a register written once and forgotten is not a risk assessment. Review it when something changes, and at least every year.
- **Trusting "we're too small to be a target"** — most attacks are automated and opportunistic. They scan for easy targets of any size.
- **Giving everyone admin rights for convenience** — it saves a few minutes and turns every mistake into a possible disaster.

## Security considerations
Security work is itself a risk area, so a few habits apply from day one:

- **Stay inside your authority.** Test, scan or probe only systems you own or have written permission to test. Curiosity is not authorisation, and unauthorised access is a crime in most countries.
- **Beware of false confidence.** A risk score looks precise but rests on guesses. Write down your assumptions and revisit them after incidents and near misses.
- **Balance security and usability.** Controls that people find unbearable get bypassed, such as passwords on sticky notes. A good control is one people will actually use.
- **People are part of the system.** Phishing and mistakes exploit human behaviour. Training helps but doesn't replace technical controls; assume that someone, someday, will click.
- **Plan for failure.** Assume a control will fail one day and ask what happens next: do you detect it, and can you recover? Test backups by restoring them.
- **Protect the register itself.** A list of your weaknesses and gaps is valuable to an attacker. Share it only with the people who need it.

## Summary
- The CIA triad names what security protects: confidentiality (only the right people see it), integrity (it is accurate and unaltered) and availability (it is there when needed). Authenticity and non-repudiation are related properties.
- Assets are what you protect, threats can cause harm, vulnerabilities are weaknesses, and risk is the chance of harm combined with its impact. Threat actors include criminals, insiders, hacktivists and nation states.
- A simple risk score is likelihood × impact on a 1–5 scale. Rank risks in a register, then mitigate, transfer, avoid or accept each one. Residual risk is never zero.
- Controls are classified by purpose (preventive, detective, corrective) and by type (technical, administrative, physical). Use a mix.
- Defence in depth uses independent layers so one failure isn't fatal; least privilege gives only the access that is needed.
- NIST CSF 2.0 groups outcomes into Govern, Identify, Protect, Detect, Respond and Recover.

## Glossary
| Term | Definition |
|---|---|
| CIA triad | The three classic security properties: confidentiality, integrity and availability. |
| Confidentiality | Information is available only to those allowed to see it. |
| Integrity | Information and systems are accurate and changed only by those allowed to change them. |
| Availability | Systems and data are usable when they are needed. |
| Data breach | An incident in which information is seen, copied or taken by someone who shouldn't have it (a confidentiality failure). |
| Distributed denial of service (DDoS) | An attack that floods a service with traffic from many sources so that real users can't use it (an availability failure). |
| Authenticity | Confidence that a person, message or file really is what it claims to be. |
| Non-repudiation | Reliable evidence, such as logs or digital signatures, that stops someone credibly denying an action. |
| Asset | Anything of value that you want to protect. |
| Threat | Anything that could cause harm to an asset. |
| Vulnerability | A weakness that a threat can use. |
| Risk | The chance that a threat uses a vulnerability to harm an asset, combined with how bad the harm would be. |
| Threat actor | A person or group that can cause harm on purpose, such as a criminal, insider, hacktivist or nation state. |
| Criminals | Threat actors motivated mainly by money. |
| Insiders | People with legitimate access, such as staff or contractors, who misuse it or make costly mistakes. |
| Hacktivists | Threat actors who attack to promote a political or social cause. |
| Nation states | Government-backed threat actors with large resources, usually after espionage or disruption. |
| Phishing | A deceptive message that tricks someone into revealing secrets, opening harmful files or paying a fake invoice. |
| Ransomware | Malware that encrypts files and demands payment for the key. |
| Misconfiguration | A system set up in an unsafe way, such as a public storage bucket or an unchanged default password. |
| Likelihood | How probable it is that a risk event happens, scored 1 to 5 in this course. |
| Impact | How severe the harm would be if it happened, scored 1 to 5 in this course. |
| Risk score | Likelihood × impact, from 1 to 25. |
| Risk matrix | A grid that maps likelihood and impact to a risk band (low, medium, high). |
| Risk register | A table of identified risks with scores, owners and treatments. |
| Qualitative / quantitative assessment | Rating risks with simple scales (qualitative) or estimating them in money (quantitative). |
| Risk treatment | The decision about what to do with a risk: mitigate, transfer, avoid or accept. |
| Mitigate | Reduce a risk by adding controls. |
| Transfer | Shift part of the financial impact of a risk to another party, for example through insurance. |
| Avoid | Stop the activity that creates a risk. |
| Accept | Knowingly live with a risk. |
| Residual risk | The risk that remains after controls are applied. |
| Security control | A safeguard that reduces a risk. |
| Preventive control | A control that stops an incident from happening. |
| Detective control | A control that notices an incident happening or having happened. |
| Corrective control | A control that limits damage and restores normal operation. |
| Deterrent control | A control that discourages people from trying, such as a warning sign. |
| Compensating control | An alternative control used when the ideal one isn't possible. |
| Technical control | A control built from software or hardware, also called a logical control. |
| Administrative control | A control made of policies, procedures and training. |
| Physical control | A control that protects the physical world, such as locks, badges and CCTV. |
| Defence in depth | Using several independent layers of controls so one failure isn't fatal. |
| Least privilege | Giving each person or system only the access needed for its job, for only as long as needed. |
| Separation of duties | Splitting a critical task between people so that no one can complete it alone. |
| NIST CSF 2.0 | The NIST Cybersecurity Framework, version 2.0 (2024): Govern, Identify, Protect, Detect, Respond, Recover. |
| Multi-factor authentication (MFA) | Logging in with two or more kinds of proof, for example a password plus a code from a phone. |

## Check your understanding
1. Give one example of a failure of each CIA property for an online shop. Which one hurts the shop most, and why?
2. A laptop with an unencrypted disk is left on a train. Name the asset, the threat, the vulnerability and the risk.
3. Two risks both score 12: one is 3 × 4 and the other 4 × 3. What does each mean, and would you treat them the same way?
4. Choose one high risk in your own life or workplace. Which treatment would you pick (mitigate, transfer, avoid or accept), and why?
5. Name one preventive, one detective and one corrective control for ransomware.
6. What is the difference between defence in depth and simply buying many security products?

## References
- [NIST Cybersecurity Framework (CSF) 2.0](https://www.nist.gov/cyberframework), 2024
- [NIST SP 800-30 Rev. 1](https://csrc.nist.gov/pubs/sp/800/30/r1/final): Guide for Conducting Risk Assessments
- [NIST SP 800-53 Rev. 5](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final): Security and Privacy Controls (a catalogue of controls)
- [FIPS 199](https://csrc.nist.gov/pubs/fips/199/final): Standards for Security Categorization (confidentiality, integrity, availability)
- [ISO/IEC 27001](https://www.iso.org/standard/27001): Information security management systems
