Skip to main content

Confidentiality and Privacy

Protecting sensitive information and respecting individual privacy rights are foundational obligations for every organization. For CPAs, these concepts are central to SOC 2® engagements, IT risk assessments, and advisory services — particularly as data breaches grow in frequency and regulatory penalties intensify. Whether you are evaluating encryption controls, assessing a client's DLP program, or testing confidentiality commitments under the Trust Services Criteria, you must understand how organizations safeguard data throughout its lifecycle. This section covers encryption fundamentals and techniques, the distinction between confidentiality and privacy, data protection methods (obfuscation, tokenization, anonymization), Data Loss Prevention (DLP), financial and operational implications of a data breach, data lifecycle controls, SOC 2® confidentiality and privacy criteria, and walkthrough procedures for confidentiality and privacy controls.

info

The ISC exam tests confidentiality and privacy at the Remembering & Understanding, Application, and Analysis levels. You should be able to explain encryption techniques, distinguish confidentiality from privacy, identify data protection methods, determine appropriate controls for the data lifecycle, and detect deficiencies in SOC 2® engagements.


Encryption Fundamentals

Encryption transforms readable data (plaintext) into an unreadable format (ciphertext) using an algorithm and a key. Only authorized parties with the correct key can decrypt the data back to plaintext.

Symmetric vs. Asymmetric Encryption

FeatureSymmetric EncryptionAsymmetric Encryption
Keys usedSingle shared key for encryption and decryptionKey pair — public key encrypts, private key decrypts
SpeedFast; efficient for large data volumesSlower; computationally intensive
Key distributionChallenge — must securely share the keyEasier — public key can be freely distributed
Common algorithmsAES, DES, 3DESRSA, ECC
Typical use casesEncrypting data at rest, bulk data transferDigital signatures, key exchange, email encryption
Exam Tip

Remember: Symmetric = Same key (one key shared by both parties). Asymmetric = A pair (public + private). AES-256 is the current industry standard for symmetric encryption. RSA and ECC are the most common asymmetric algorithms.

Hashing

Hashing is a one-way function that produces a fixed-length output (hash or digest) from any input. Unlike encryption, hashing cannot be reversed. It verifies data integrity rather than providing confidentiality.

AlgorithmOutput LengthStatus
MD5128 bitsDeprecated — collision vulnerabilities
SHA-1160 bitsDeprecated — collision vulnerabilities
SHA-256256 bitsCurrent standard
SHA-3VariableNewest standard

Public Key Infrastructure (PKI) and Digital Certificates

PKI provides the framework for managing digital certificates and public-key encryption. Components include:

  • Certificate Authority (CA) — Issues and signs digital certificates
  • Registration Authority (RA) — Verifies identity before certificate issuance
  • Digital certificate — Binds a public key to an entity's identity
  • Certificate Revocation List (CRL) — Lists revoked certificates

Encryption at Rest vs. In Transit

StateDescriptionCommon Controls
At restData stored on disk, database, or backup mediaAES-256 full-disk encryption, database TDE
In transitData moving across a networkTLS 1.2+, IPsec VPN, HTTPS

TLS (Transport Layer Security) replaced the deprecated SSL protocol and secures data in transit by establishing an encrypted channel between two endpoints using both asymmetric (for key exchange) and symmetric (for bulk data) encryption.

Confidentiality vs. Privacy

Although related, confidentiality and privacy address different concerns:

DimensionConfidentialityPrivacy
DefinitionProtecting sensitive information from unauthorized access or disclosureAn individual's right to control the collection, use, and sharing of their personal information
FocusThe data itself — trade secrets, financial records, IPThe data subject — customers, employees, patients
ObligationContractual or organizational (e.g., NDA, policy)Legal and regulatory (e.g., GDPR, HIPAA, CCPA)
TSC categoryConfidentiality (C1)Privacy (P1–P8)
ExamplePreventing unauthorized access to Bear Co.'s merger plansEnsuring Bear Co. obtains consent before collecting customer emails
note

All privacy protections require confidentiality controls, but not all confidential information involves personal data. A company's proprietary formula is confidential but does not trigger privacy regulations.


Data Protection Methods

Organizations employ various techniques to protect confidential data during the design, development, testing, and implementation of applications.

MethodDescriptionReversible?Use Case
Data maskingReplaces sensitive characters with substitutes (e.g., XXX-XX-1234)No (static) / Yes (dynamic)Displaying partial data in UIs
Data scramblingRandomly rearranges characters within a fieldYes (with key)Non-production test environments
TokenizationReplaces sensitive data with a non-sensitive token; original stored in a secure vaultYes (via vault lookup)Payment card processing (PCI DSS)
AnonymizationIrreversibly removes identifying informationNoResearch datasets, analytics
PseudonymizationReplaces identifiers with artificial ones; re-identification possible with a separate keyYes (with key)GDPR-compliant processing

When to Apply During the SDLC

Example: MAS Inc. builds a new payroll application. During testing, developers must never use real employee SSNs. The security team provides tokenized test data so that the application logic can be validated without exposing actual sensitive values.

Data Loss Prevention (DLP)

Data Loss Prevention (DLP) refers to the set of tools, processes, and policies that detect and prevent unauthorized transmission or leakage of sensitive data outside the organization.

Types of DLP

TypeDeploymentWhat It Monitors
Network DLPMonitors network traffic at egress pointsEmail, web uploads, FTP transfers
Endpoint DLPInstalled on workstations and laptopsUSB copies, print jobs, clipboard actions
Cloud DLPIntegrates with SaaS/IaaS platformsCloud storage sharing, API data flows

How DLP Works

  1. Content inspection — Scans data for patterns (credit card numbers, SSNs, keywords)
  2. Contextual analysis — Evaluates sender, recipient, time, destination
  3. Policy enforcement — Blocks, quarantines, encrypts, or alerts based on rule matches
    warning

    DLP is only effective when policies are kept current and aligned with data classification. A DLP system that cannot identify newly classified confidential data will fail to prevent its exfiltration.

    Example: Illini Security configures its network DLP to flag any outbound email containing more than 10 records matching a SSN pattern (XXX-XX-XXXX). When an employee attempts to email a spreadsheet with 500 customer SSNs to a personal account, the DLP system blocks the transmission and alerts the security team.

Financial and Operational Implications of a Data Breach

Impact CategoryExamplesTypical Cost Drivers
FinancialRegulatory fines, legal settlements, credit monitoring, forensic investigationGDPR fines up to €20M or 4% of global revenue; HIPAA fines up to $1.5M per violation category
OperationalBusiness disruption, system downtime, diverted IT resourcesIncident response labor, emergency patching, overtime
ReputationalLoss of customer trust, negative media coverage, stock price declineCustomer churn, reduced new business, investor confidence
Legal/RegulatoryLawsuits, consent decrees, mandatory auditsAttorney fees, ongoing compliance monitoring
InsuranceCyber liability premiums increase, coverage disputesPolicy deductibles, exclusions for negligence
note

The average cost of a data breach exceeds $4 million globally. Organizations with mature encryption and DLP programs experience significantly lower breach costs due to reduced data exposure.

Example: Kingfisher Industries suffers a breach exposing 200,000 customer payment records. Direct costs include $2M in notification and credit monitoring, $1.5M in forensic investigation, and a $3M regulatory fine. Indirect costs — customer attrition and brand damage — exceed $5M over two years.


Data Lifecycle Controls

Organizations must implement controls at every stage of the data lifecycle:

StageKey Controls
CollectionConsent mechanisms, data minimization, input validation, classification at point of entry
ProcessingRole-based access control (RBAC), least privilege, audit logging, data quality checks
StorageEncryption at rest (AES-256), access control lists, data classification labels, backup encryption
TransmissionTLS 1.2+, VPN tunnels, encrypted email (S/MIME or PGP), secure file transfer (SFTP)
DeletionCrypto-shredding, secure wipe (NIST 800-88), certificate of destruction, retention policy enforcement

Data Classification

LevelDescriptionExample
PublicNo harm if disclosedMarketing materials
InternalMinor harm if disclosedInternal memos
ConfidentialSignificant harm if disclosedFinancial statements (pre-release)
RestrictedSevere harm if disclosedPII, PHI, trade secrets

Secure Destruction Methods

  • Crypto-shredding — Destroying encryption keys renders data unrecoverable without physically destroying media
  • Degaussing — Magnetically erasing data from magnetic media
  • Physical destruction — Shredding, incineration, or disintegration of storage media
  • Secure overwrite — Writing random patterns over data multiple times (per NIST SP 800-88)

SOC 2® Confidentiality and Privacy Controls

In a SOC 2® engagement, auditors evaluate an organization's controls against the Trust Services Criteria (TSC). Confidentiality and privacy are two of the five TSC categories.

Confidentiality Criteria (C1)

The confidentiality category evaluates whether the entity protects information designated as confidential:

CriterionFocus Area
C1.1Confidential information is identified and protected
C1.2Confidential information is disposed of securely

Privacy Criteria (P1–P8)

CriterionPrincipleDescription
P1NoticeThe entity provides notice about its privacy practices
P2Choice and consentThe entity obtains consent for collection and use of personal information
P3CollectionPersonal information is collected only for identified purposes
P4Use, retention, and disposalPersonal information is used, retained, and disposed of per stated policies
P5AccessData subjects can access and update their personal information
P6Disclosure and notificationPersonal information is disclosed only as authorized; breach notification is provided
P7QualityThe entity maintains accurate, complete personal information
P8Monitoring and enforcementThe entity monitors compliance with its privacy commitments

Design Deficiencies vs. Operating Deviations

Issue TypeDefinitionExample
Design deficiencyA control is missing or improperly designed such that even if it operates as intended, it cannot meet the criterionGies Co. has no policy requiring encryption of confidential data at rest — the control simply does not exist
Operating deviationA properly designed control fails to operate as intended during the examination periodGies Co. has an encryption policy but an administrator bypassed it for three months due to a system migration

Confidentiality and Privacy Walkthroughs

A walkthrough is a procedure in which the auditor traces a transaction or process from initiation to completion, observing actual procedures and comparing them to documented policies.

Walkthrough Procedure Steps

  1. Select the process — Choose a confidentiality or privacy-relevant process (e.g., customer data collection, data disposal)
  2. Obtain documentation — Gather written policies and procedure documents
  3. Trace through the system — Follow data from input through processing to output/destruction
  4. Observe personnel — Watch employees perform the process in real time
  5. Inquire — Ask operators clarifying questions about exceptions and controls
  6. Compare — Evaluate observed procedure against documented requirements

Areas to Evaluate

AreaWhat to Look For
IT risk managementRisk assessments include confidentiality and privacy threats
Human resourcesBackground checks, termination procedures revoke access promptly
Training and educationEmployees receive privacy awareness training annually
Incident responseBreach notification procedures align with regulatory timelines
Vendor managementThird-party contracts include confidentiality and data protection clauses

Example: Illini Entertainment documents a policy requiring employee access to customer PII to be revoked within 24 hours of termination. During a walkthrough, the auditor discovers that two terminated employees retained access for 15 days. This is an operating deviation — the policy is properly designed but was not consistently executed.

Summary

TopicKey Takeaway
Encryption fundamentalsSymmetric (AES) is fast for bulk data; asymmetric (RSA/ECC) enables secure key exchange; hashing verifies integrity
Confidentiality vs. privacyConfidentiality protects data from unauthorized access; privacy protects individuals' rights over their personal data
Data protection methodsTokenization, masking, anonymization, and pseudonymization protect data during SDLC — choose based on reversibility needs
Data Loss PreventionDLP tools monitor network, endpoint, and cloud channels using content inspection and contextual analysis to prevent data leakage
Data breach implicationsBreaches cause financial (fines, remediation), operational (downtime), reputational (churn), and legal consequences
Data lifecycle controlsApply classification, encryption, access controls, and secure destruction at each stage — collection through deletion
SOC 2® criteriaConfidentiality (C1) and Privacy (P1–P8) criteria define the benchmarks; distinguish design deficiencies from operating deviations
WalkthroughsTrace processes end-to-end, observe personnel, and compare actual practices to documented policies

Practice Questions

  1. Bear Co. stores customer payment data in a cloud database encrypted with AES-256. When transmitting this data to a payment processor, the connection uses TLS 1.3. A new developer proposes using MD5 hashes to verify data integrity during transfer. Should the security team approve this approach, and why or why not?
  2. MAS Inc. is undergoing a SOC 2® engagement. The auditor reviews the company's confidentiality controls and finds that the data retention policy requires deletion of client files 90 days after engagement completion. However, when the auditor examines the file server, client files from 8 months ago remain accessible. Is this a design deficiency or an operating deviation? Explain.
  3. Illini Security implements a new DLP solution. During configuration, the team sets rules to block outbound emails containing credit card numbers but does not create rules for Social Security numbers, health records, or proprietary source code. Six months later, an employee emails a file containing 1,000 patient health records to an unauthorized external recipient without triggering any alert. What type of DLP failure is this, and what should the team have done differently?
    Answers
  4. The security team should reject MD5 for integrity verification. MD5 is cryptographically broken — known collision vulnerabilities allow attackers to craft two different inputs with the same hash. Bear Co. should use SHA-256 or higher for integrity checks. While TLS 1.3 already provides message integrity via its built-in HMAC, any additional application-layer integrity check must use a current algorithm.
  5. This is an operating deviation, not a design deficiency. The control (90-day retention policy with deletion requirement) is properly designed and, if followed, would satisfy the confidentiality criterion. The failure is in execution — personnel did not perform the scheduled deletion. The auditor should report this as a deviation in control operation during the examination period.
  6. This is a policy coverage failure — the DLP rules were incomplete. DLP effectiveness depends on comprehensive policies aligned with the organization's data classification scheme. Illini Security should have: (a) mapped all categories of sensitive data (PCI, PII, PHI, IP) to DLP rules, (b) conducted a data classification exercise before configuring rules, and (c) tested detection patterns for each data category. A DLP system can only prevent leakage of data types it is configured to recognize. :::