Skip to content
Tutorials Center
My learning

Topics

Lesson 4 of 53 ยท Foundations and Ethics

Classic Security Models

Learn the classic access control and security models: Bell-LaPadula for confidentiality, Biba for integrity, Clark-Wilson for transactions, plus how they show up in real systems.

Before you can break access control, it helps to know what a correct design looks like. Security researchers formalized the idea decades ago in security models: precise sets of rules that say who may read or write what, so that a system can be reasoned about and checked. You will rarely implement these models literally, but their vocabulary and their ideas appear everywhere: classification labels, role-based access, write-up and read-down rules, audit trails and separation of duties.

This lesson covers three classic models, how they differ and where their ideas live in modern systems. It builds on the CIA triad from Introduction to Ethical Hacking and connects to Threat Modeling. The code examples are small simulations of the rules.

Subjects, objects and labels

Every model starts with the same cast.

  • A subject is an active entity: a user or process.
  • An object is a passive resource: a file, record or database table.
  • A label or level classifies both. Subjects get a clearance; objects get a classification. Levels are ordered, for example Public < Internal < Confidential < Secret.

A model then says which operations (read, write) are allowed between a subject and an object, based on those levels. Two kinds of control are common. Mandatory access control (MAC) is enforced by the system according to labels; users cannot override it. Discretionary access control (DAC) lets owners decide who gets access, as with ordinary file permissions.

Bell-LaPadula: protecting confidentiality

The Bell-LaPadula (BLP) model was designed for military-style secrecy. It aims to stop information from leaking to people who should not see it. It has two key rules:

  • Simple security property ("no read up"): a subject may not read an object at a higher level than its clearance. A person with Confidential clearance cannot read Secret documents.
  • *-property ("no write down"): a subject may not write to an object at a lower level. Someone with Secret clearance cannot save content into a Public file, because that could leak secrets downward, intentionally or by accident.

BLP cares about confidentiality only. It says nothing about whether data is correct. Its "write down" rule is the reason that systems with classifications often forbid copying sensitive data into low-sensitivity places, which you see today in data loss prevention (DLP) policies.

Biba: protecting integrity

The Biba model flips Bell-LaPadula to protect integrity: it prevents unreliable data from corrupting trustworthy data. Its rules are:

  • Simple integrity property ("no read down"): a subject may not read data at a lower integrity level, because bad data could contaminate its work.
  • *-integrity property ("no write up"): a subject may not write to an object at a higher integrity level, so a low-trust process cannot alter high-trust data.

In practice: a trusted configuration file should not be modifiable by an untrusted program, and a critical process should be careful about consuming unverified input. Operating systems apply this idea with integrity levels for processes (a browser tab running at low integrity cannot overwrite normal user files) and with sandboxing.

Clark-Wilson: integrity for business

Real organizations care about integrity in a different way than a flat "no write up" rule. The Clark-Wilson model is built around business transactions and has three parts:

  • Constrained data items (CDIs): the important data, such as account balances, that must stay consistent.
  • Transformation procedures (TPs): the only approved ways to change a CDI, for example "transfer funds", which keep the data valid (debits equal credits).
  • Integrity verification procedures (IVPs): checks that confirm the data is in a valid state.

Users do not touch the data directly. They are allowed to run specific TPs, which enforces well-formed transactions and separation of duties: the person who approves a payment is not the person who creates it. Everything is logged to an audit trail. Clark-Wilson is the ancestor of the controls in accounting software, change management and any system where the interface is "you may do these specific actions", not "you may edit these tables". Compare this to the business logic flaws that arise when a TP does not enforce its rules.

Three models at a glance

Bell-LaPadula Confidentiality. No read up, no write down.
Biba Integrity. No read down, no write up.
Clark-Wilson Integrity by approved transactions.

Different goals, different rules.

1 / 2

Simulating the rules

The stand-in below implements Bell-LaPadula and Biba as functions and tests a few access requests. Seeing the two models answer the same request differently makes the contrast clear.

Node.js Bell-LaPadula and Biba as functions

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with node:readline or process.stdin), it gets the next line from this box.

Read the table row by row. A Secret subject reading Internal data is fine for confidentiality (no leak) but violates Biba (it might trust low-quality data). A low-level subject writing up is fine for BLP (information moves upward) but violates Biba (it could corrupt high-integrity data).

Simulating Clark-Wilson

The second stand-in has a balance as the CDI, a single transformation procedure that enforces rules, an integrity check and a separation of duties condition. Direct edits are not possible; only the TP can change the data.

Node.js A Clark-Wilson style transaction

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with node:readline or process.stdin), it gets the next line from this box.

The negative-amount rejection is the lesson: without the TP's checks, a "transfer" of -500 would steal money. That is a classic business logic bug.

Where the models appear today

  • Classification labels on documents and email (Public, Internal, Confidential) and the rules that stop sending Confidential data to personal accounts echo Bell-LaPadula.
  • Integrity levels, sandboxing and signed updates echo Biba: untrusted code must not alter trusted code.
  • Role-based access control, approval workflows, audit logs echo Clark-Wilson.
  • SELinux and AppArmor implement mandatory access control on Linux.
  • Cloud policies and least privilege are practical descendants.

Models are also limited. BLP ignores integrity and covert channels; Biba ignores confidentiality; real systems usually need a combination. Treat the models as thinking tools for spotting what a design forgot.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. In Bell-LaPadula, a user with Confidential clearance tries to read a Secret document. What happens?

    The simple security property forbids reading at a higher level than your clearance.
  2. 2. Which model is primarily concerned with preventing low-trust data from corrupting high-trust data?

    Biba protects integrity with 'no read down' and 'no write up' rules.
  3. 3. What are transformation procedures in Clark-Wilson?

    Users act through transformation procedures, which keep the data valid and enforce rules such as separation of duties.
  4. 4. Which real-world mechanism is closest to the Clark-Wilson idea?

    Constrained actions, separation of duties and audit trails are the core of Clark-Wilson.

Summary

  • Security models define who may read and write which objects, based on subjects, objects and labels.
  • Bell-LaPadula protects confidentiality: no read up, no write down.
  • Biba protects integrity: no read down, no write up.
  • Clark-Wilson protects integrity through approved transactions, separation of duties and audit trails.
  • Mandatory access control enforces rules by system policy; discretionary control leaves decisions to owners.
  • Models are thinking tools: use them to ask what a design forgot.

Next: The Penetration Testing Process