Confidentiality
Making sure only the right people can see a piece of information.
Draft - this entry has not been reviewed yet.
Formal
The property that information is not made available or disclosed to people, systems or processes that are not allowed to have it - whether it is stored, being sent or being printed.
In plain English
Like a sealed letter - the address is on the outside, but only the person it is meant for should read what is inside.
In practice
In a municipality's HR system, salary details can be seen only by the payroll team and each employee's own manager, not by colleagues in other departments.
Why it matters
Once secret information has leaked it cannot be taken back, and a leak of personal data must be reported to the authorities and can bring fines and lasting damage to trust.
Technical deep dive
ISO/IEC 27000 defines confidentiality as the property that information is not made available or disclosed to unauthorised individuals, entities or processes. It is enforced by two families of mechanism that are easy to confuse. Access control decides who may read an object while it sits inside a system that enforces the rules; encryption protects the object when it leaves that enforcement boundary, on the wire, on a stolen disk, in a backup or at a cloud provider. Neither replaces the other: an encrypted database is fully readable by an application account with excessive privileges, and perfect access control is irrelevant once a file has been copied to a USB stick.
Access-control models formalise the "who". Discretionary access control (DAC) lets owners grant access, as with file-system ACLs; mandatory access control (MAC) enforces system-wide labels, the classic example being the Bell-LaPadula model (1973) with its "no read up, no write down" rules for classified data; role-based access control (RBAC) attaches permissions to job roles; attribute-based access control (ABAC) evaluates policies over user, resource and context attributes at request time. Across all of them the governing principles are need-to-know and least privilege, and the operational weak point is usually privilege creep as people change jobs.
For encryption, the usual split is data in transit (TLS 1.3, RFC 8446, with forward secrecy through ephemeral Diffie-Hellman so that a later key compromise does not expose recorded sessions), data at rest (full-disk, database or object-level encryption, where the real question is who controls the keys) and, increasingly, data in use through confidential computing enclaves. "Harvest now, decrypt later" is a confidentiality threat specific to long-lived secrets, and is the main argument for migrating to post-quantum key exchange before large quantum computers exist.
Confidentiality also leaks through channels that access control does not model: metadata such as who communicates with whom and when, traffic analysis of encrypted flows, error messages and timing differences, side channels such as the Spectre and Meltdown CPU flaws disclosed in 2018, and inference from aggregated or "anonymised" datasets. Confidentiality is distinct from privacy, which is about lawful and fair processing of personal data even by authorised parties, and from integrity, which concerns modification rather than disclosure. Under GDPR, Art. 5(1)(f) and Art. 32 require appropriate confidentiality measures, with Art. 32(1)(a) naming encryption and pseudonymisation, and unauthorised disclosure is a personal data breach that must be assessed for notification to Datatilsynet within 72 hours under Art. 33.
Relationships
- Part of
- CIA triad
- Implemented by
- EncryptionEnd-to-end encryptionHTTPSTLSVPN
- Don't confuse with
- IntegrityAvailability
Sources & further reading
Standards & official texts
- ISO/IEC 27000:2018 - Information security management systems - Overview and vocabulary · ISO/IEC
Course material
- Cyber Security Fast Track - Ordliste
Where this data comes from
This entry was drafted by an AI from the sources above and has not yet been checked by a person. Treat it as a starting point, and check anything important against the sources.
See the review queueSuggest a correction on GitHubThis term as JSON
Mentioned in
Check yourself
Loading…