What we track and how we label it.

The entity is named by Anthropic or another primary source as a participant, partner, evaluator, or access channel.

A credible news source names the entity, but the entity is not treated as self-confirming in the tracker.

Public reporting describes access guardrails, procurement preparation, or a possible deployment path without confirmed use.

Public reporting says a technical or government body evaluated the model directly.

An institution has oversight relevance or public concern, but public reporting does not show direct model testing or access.

A named regulator or ministry has taken a concrete public or reported action in response to Mythos or directly adjacent cyber-AI capability.

A named regulator is monitoring, briefing, or coordinating on possible implications without evidence of direct model access.

A country or market has material Claude or AI-governance engagement, but no public Mythos-specific access record.

A sourced contextual entry on infrastructure, cyber capacity, or governance participation relevant to Mythos-class capability distribution.

A contextual record used to track absence, infrastructure alignment, or strategic relevance. It is not access evidence.

Access is denied, unavailable, restricted, or materially contested in public reporting.

A source discloses a group without naming the individual organizations.