TLP 2.0: sharing boundaries are not information classification

On 10 October, Jan Kopriva at SANS Internet Storm Center discussed the confusion between the Traffic Light Protocol and internal information classification. This is the author’s commentary on how labels are used, rather than an announcement of a new standard version. Source: the SANS analysis of TLP and classification.

Breachroad’s conclusion: teams exchanging threat reports should check permitted recipients and document handling requirements separately. This applies to analysts, administrators, business partners and people using AI assistants. A label’s colour does not answer every question about working with information.

What the labels mean

FIRST’s TLP 2.0 defines sharing boundaries; it does not replace formal classification or encryption rules.

Label	Sharing boundary
TLP:RED	Named individual recipients only.
TLP:AMBER	The organisation and its clients, on a need-to-know basis.
TLP:AMBER+STRICT	The organisation only, on a need-to-know basis.
TLP:GREEN	A defined community, excluding public channels.
TLP:CLEAR	No TLP sharing restrictions, subject to copyright.

Wider sharing requires permission from the source. If a GREEN community is undefined, the default is the cybersecurity defence community. Definition source: the official FIRST TLP 2.0 standard.

Two questions before forwarding a report

The following workflow is our recommendation, rather than an additional TLP label. First establish whether the intended recipient falls within the permitted audience. Identify the person, organisation or group and any further instructions from the sender. An existing working relationship does not replace that assessment.

Then establish whether your organisation permits the proposed tool and processing method. Will the document enter an approved workspace? Who can access it? Will it contain personal data, architecture details or customer identifiers? These questions concern the actual material, rather than its colour alone.

We suggest recording both decisions in one ticket. The person carrying out the task should see the permitted audience, storage location, sharing purpose and an owner for outstanding questions. That allows another team to understand the decision without reconstructing it from a chat conversation.

A partner asks for the full report

A team receives a report with a restricted audience. A project partner requests a complete copy to understand its recommendations. This is a hypothetical training scenario. Reaching an appropriate decision requires checking the sharing conditions, rather than assuming that a commercial relationship automatically satisfies them.

Our recommendation is to identify the report owner and establish what the partner needs. If the proposed disclosure exceeds the authorisation, obtain a decision from the source before sending it. Do not assume that a summary removes the restrictions: it may still contain the same protected details. An agreed, limited extract may be sufficient, but its disclosure still needs assessment.

During the exercise, ask participants for a short explanation of their decision. It should state which information was available and what remained unresolved. Naming a colour alone does not show whether someone can apply the rule to a real recipient.

An employee wants an AI summary

Another employee wants to paste the same report into an AI tool. In our assessment, that requires a separate review of the service, account, data scope and organisational rules. A TLP label should not be treated as automatic permission to process a report in any application.

We recommend providing an approved method for producing these summaries, with a clear reporting route when a tool or document raises questions. Employees need a practical answer about where they can complete the task and who decides an exception. A list of prohibitions alone does not teach that decision.

The training discussion can use a fictional report and a fictional workspace. Participants can identify the proposed destination, the data it would receive and the missing decision without uploading any sensitive material. This is our suggested exercise design, rather than a requirement in the standard.

What to practise with your team

Compare three decisions in a training session: sharing with a partner, publishing an extract and summarising in an approved tool. Ask participants to identify the recipient, purpose and any missing permission or information. This is our recommended knowledge exercise format; it does not require real confidential documents.

For broader context on using reports, read our guide to the threat intelligence lifecycle. To improve decisions about information sharing and AI use, explore our cybersecurity training for organisations.