Classification

In the context of information classification, a cryptographic key is often viewed merely as another technical safeguard. However, this is a somewhat misleading simplification. A key does not simply protect information from unauthorized reading; it can also enable the creation of signatures, verify data origin, authenticate users or systems, and provide time-stamping services. Consequently, the compromise of such a key can have vastly different repercussions compared to the compromise of a key used for standard data encryption.
Therefore, the classification of key material cannot rely solely on the sensitivity of the data the key protects. It is more important to determine what actions an attacker could take following a key compromise, which entities or systems could be affected, and the duration of the resulting impact.
This also ties into the previous part of the series: a key is a security asset with its own purpose, owner, dependencies, and lifecycle. Classification is one of the parameters used to subsequently determine the appropriate protection measures.

Classification of Key Material

Key material is used to perform cryptographic operations that protect information or ensure its integrity and authenticity. As such, the key material itself can be a highly sensitive asset, and the measures taken to protect it must be commensurate with the consequences of its compromise. The situation is relatively easy to visualize in the case of encryption keys. If a key protects data of a certain sensitivity level, its compromise could lead to that data being exposed. Therefore, as an organizational rule, it makes sense to require that the level of protection for the key corresponds at least to the sensitivity of the information it protects.
However, this rule cannot be applied mechanically to all types of keys. A key does not necessarily protect only confidentiality; its compromise could also undermine the integrity, authenticity, or evidentiary value of cryptographic operations. Consequently, classification should be based primarily on the most serious realistic scenario of misuse. In other words, a key is not sensitive simply because it "encrypts secret data"; it is also sensitive because its holder gains the ability to create, alter, or confirm something, or to impersonate another entity.

Signing keys

The situation regarding signing keys differs from that of encryption keys. Compromising a signing key does not necessarily result in an information leak. However, an attacker could gain the ability to generate new signatures on behalf of the key holder. The impact therefore depends not only on the type of key itself but also on the privileges of the entity to which the key is assigned. A signing key belonging to a person authorized to approve significant financial or legal transactions may represent a far more sensitive asset than a key belonging to a person without such authorization. For signing keys, it is therefore advisable to monitor, in particular, the relationship:
identity – key – authorization – signing operation

If a key is compromised, simply replacing the key technically is insufficient. It is necessary to determine what operations the attacker could have performed, what signatures could have been created, and what impact the questioning of their origin might have.
The concept of non-repudiation deserves special attention. A cryptographic signature in itself does not automatically equate to legal non-repudiation. While from a technical standpoint it guarantees document integrity (meaning the text has not been altered since the moment of signing), from a legal perspective it serves as key evidence of the signatory's intent. Cryptography thus provides only a technical mechanism for verifying the signature and its link to a specific key. However, the actual evidentiary and legal effects depend on the identification of the holder, the method of key protection, authorization management, the overall chain of evidence, and the applicable legal regime. It is precisely based on these factors that a court or authority interprets the law and determines the resulting non-repudiation of the legal act in question.
Therefore, it is not appropriate to classify a signing key solely based on the sensitivity of the documents being signed. One must also consider the legal or operational actions the key holder can perform on behalf of the entity in question.

Sealing keys

A sealing key has a different usage model. An electronic seal is associated with a legal entity, and its purpose is primarily to demonstrate the origin and integrity of data. It does not, therefore, represent the expression of a specific natural person's will in the same way as a personal electronic signature.
Consequently, the compromise of a sealing key does not necessarily lead to an information leak. However, an attacker could gain the ability to generate data that appears to bear the organization's seal. Consequently, the impact could extend to items such as invoices, purchase orders, statements, certificates, or signed software. Therefore, the classification of a sealing key should take into account not only the technical protection of the key but also the level of trust associated with its use.
For sealing keys, the following relationship is particularly important:
organization – key – usage authorization – seal – data trustworthiness

TSA Keys

A key used by a Time Stamping Authority (TSA) service has a different nature. Its purpose is neither to protect data confidentiality nor to confirm the intent of a specific individual. It is used to generate time stamps that provide an evidentiary link between specific data and a particular point in time. More precisely, a time stamp provides proof that specific data was submitted to the service in a given form no later than the time indicated in the time stamp. The time stamp itself does not prove who created the document, who owned it, or who signed it. These facts must be substantiated by additional mechanisms.
Compromise of a TSA key can therefore undermine the trustworthiness of the entire service. It is not merely a matter of compromising a single document or a single user. The issue can have broader implications for timestamps issued by the service and for the chains of evidence that rely on them. Consequently, the classification of a TSA key must take into account not only the key itself but also the scope of the service and the volume of systems and documents that depend on its trustworthiness.

Authentication keys

An authentication key is typically not used for data encryption. Its purpose is to verify the identity of a user, device, or service. Compromise of such a key can allow an attacker to impersonate a specific entity. Subsequently, the attacker could gain access to systems and data commensurate with that entity's privileges. Therefore, when classifying an authentication key, it is crucial to consider the scope of privileges associated with the authenticated identity. Simply put:
compromise of authentication key → compromise of identity → potential exploitation of identity privileges

Thus, an authentication key cannot be classified solely based on whether it "protects data." It is necessary to assess the capabilities an attacker would gain upon its compromise.

Classification based on the consequences of compromise

A common principle emerges from the examples provided. The key type is important, but it does not determine the key's sensitivity on its own. When classifying keys, it is useful to ask the following questions in particular:

• What is the purpose of the key?
• What security objective does its use support?
• What authorization is associated with its use?
• Which entities, systems, or information rely on the key?
• What actions can an attacker perform if the key is compromised?
• How extensive could the consequences of a compromise be?
• How long might the consequences persist?
• Can a compromise be detected, and how quickly can the key be replaced?
• Are there evidentiary or legal implications regarding operations already performed?

This approach also makes it possible to distinguish between two keys of the same technical type that may require vastly different levels of protection. For instance, two encryption keys might use the same algorithm and have the same length, yet one might protect temporary operational data while the other secures a long-term sensitive archive. Similarly, two signing keys might utilize the same algorithm but have vastly different impacts depending on the authorization of their holders.
Classification is therefore not a property of the algorithm itself; rather, it is a property of the asset within the specific context of its use.

Sensitivity of Key Material

Cryptographic key material is a sensitive component of information systems. However, the security of a cryptographic mechanism does not rely solely on the algorithm, certificate, or communication protocol. Equally important is the administration of key material and the management of its lifecycle. Key questions include, in particular:

• who is authorized to create the key material,
• who owns it,
• who is the authorized holder,
• who is responsible for its protection,
• who may use the key,
• for what purpose it may be used,
• who may administer it,
• how the authorization of a specific cryptographic operation can be proven,
• how the key material is recorded.

However, recording does not mean recording the secret itself. As stated in the previous part of the series, an organization primarily needs to record metadata and associations that enable the key to be managed. Conversely, the private key itself should not be made accessible to a standard record-keeping system merely to claim that it is “recorded.”

Unambiguous association for personal signing keys

For personal signing keys, it is crucial to maintain an unambiguous association between:

• the holder's identity,
• the specific cryptographic key,
• the specific signing operation.

If the same key is used by multiple individuals without a mechanism that allows for the unambiguous identification of the authorized holder and their control over the key, issues regarding accountability and proof arise. It is then impossible to reliably determine who created a specific signature and whether the signing operation was performed under the holder's control. This is also why it is inadvisable to conflate key ownership, key management, and the authorization to use the key. For example, a technical administrator might manage system operations without thereby gaining the authority to sign on behalf of the key holder.

Keys protecting other keys

Keys that protect or affect other cryptographic keys deserve special attention. The compromise of a certification authority key can have a vastly different impact than the compromise of a single server certificate. This creates a hierarchy of trust that must be taken into account during both classification and asset management. The more objects that depend on a given key, the greater the potential consequences of its compromise.

The temporal aspect of classification

Classification has another dimension: the value of information may not remain constant over time constant. Some information holds high value only before or during a specific event. Once the event concludes, its immediate significance may rapidly diminish. Other information may possess long-term value for analysis, planning, or the retrospective inference of system properties.
The potential to combine multiple sources presents a particular challenge. A single data point may become virtually meaningless after a certain period, yet a large volume of similar data can enable the retrospective inference of process attributes, measurement accuracy, response times, or other system characteristics.
Consequently, the change in information value over time is not necessarily constant (or monotonic). Immediate value may drop sharply, while analytical or evidentiary value persists. Therefore, when classifying information, it is advisable to consider at least the following factors:

• the time window during which the information is sensitive,
• the rate at which its value changes,
• the potential for retrospective reconstruction,
• the potential for combination with other information,
• requirements regarding long-term confidentiality or evidentiary value.

This same line of reasoning applies to key protection. For certain encryption keys, protection is critical specifically during the period when the protected data remains sensitive. In the case of signing or time-stamping keys, however, the long-term preservation of the trustworthiness of operations already performed may be crucial.

Non-repudiation

Cryptographic and legal perspectives on non-repudiation differ, often leading to significant conflict and misunderstanding. While related, these two perspectives are not identical. A valid cryptographic signature does not automatically prove that a specific individual knowingly and voluntarily signed a document or agreed to its contents. This distinction is crucial for a proper understanding of non-repudiation.
Cryptographic perspective: It is possible to verify that the signature corresponds to a specific public key and that the signed data has not been altered.
Legal perspective: Based on available evidence, the action can be attributed to a specific person, and legal consequences can be derived from it.
But why does a digital signature not automatically guarantee non-repudiation? The reason lies in the link between the key and the specific person (the key owner). Complications arise if that person fails to adequately protect the key. There are various examples of such situations. From unsecured storage to a stolen key or a signature generated by an automated system without direct user intervention (such as verifiable authentication)—and extending to scenarios like key loss or a Certification Authority's failure to correctly verify a user's identity... all these instances highlight a weakness in the link between the user and the key. Cryptographic verification can confirm that a signature corresponds to the private key paired with a specific public key; however, without further evidence, it cannot prove that the key was used solely by the authorized user.
Legal analysis goes beyond the question of whether a signature can be mathematically verified. It also examines whether a specific action can be attributed to a particular individual and what legal consequences may arise from it. Within the European Union, these matters are governed by the eIDAS regulation; the relationship between cryptographic key material and this regulation will be discussed in a future article.

Conclusion

The classification of cryptographic key material should not be derived solely from the key's name, the algorithm used, or the classification of the data the key protects. The decisive factor is primarily the impact of its compromise. For encryption keys, this impact often depends on the confidentiality of the protected information. For signing and sealing keys, the ability to generate new cryptographic evidence on behalf of a specific person or organization is crucial. For authentication keys, it is necessary to assess the scope of privileges associated with the compromised identity. For TSA keys, the scope of the service and the impact on the trustworthiness of previously issued time stamps are significant.
At the same time, it is necessary to distinguish the technical characteristics of a cryptographic mechanism from its legal effects. A cryptographic signature or time stamp alone does not automatically determine legal evidentiary weight. That depends also on the method of subject identification, key protection, privilege management, the chain of evidence, and the applicable legal regime. Therefore, the practical outcome of the classification should not merely be a label in a registry. Classification determines the level of protection a key requires, who is authorized to use it, how it must be audited, and the procedures to be followed in the event of its compromise, replacement, or end-of-lifecycle.
Only when the key's purpose, its dependencies, and the consequences of its compromise are known can we decide how to protect it. The next part of the series addresses precisely this issue: defining the key's purpose and usage rules.

Autor článku:

Jan Dušátko

Jan Dušátko has been working with computers and computer security for almost a quarter of a century. In the field of cryptography, he has cooperated with leading experts such as Vlastimil Klíma or Tomáš Rosa. Currently he works as a security consultant, his main focus is on topics related to cryptography, security, e-mail communication and Linux systems.