Frontier-Class Designation of AI Models

Frontier-class designation

Definition

A frontier-class designation is a formal decision by a named authority that a specific AI model belongs in a class requiring special oversight, recorded with the issuing body, the criteria used and the date, rather than left to the developer's own classification.

Also called: frontier model designation, systemic-risk designation, covered frontier model designation, Frontier-class

NIKOLAI's proposed operational definition

NIKOLAI proposal (unsourced): a record that a named authority has formally designated a specific model as falling in a special-oversight class, capturing the issuing body, the legal or policy basis, the criteria applied, the date, and any route to contest or reassess the designation. It differs from coverage-scope-threshold, which is the test itself: a model can meet a compute presumption without any designation, and an authority can designate a model the developer did not self-classify.

The operational definition is CASRAI's own proposal, not a definition any named organisation has agreed to. See what NIKOLAI is and is not, or browse the AI safety glossary.

Examples

Examples from published documents

Under the EU AI Act, "The Commission may designate a general-purpose AI model as presenting systemic risks, ex officio or following a qualified alert from the scientific panel" (EU AI Act, Article 52(4))
Executive Order 14409 assigns the designation of a "covered frontier model" to the Director of NSA, after consultation (Executive Order 14409, Sec. 3(a))

Counter-examples

Not to be confused with

A developer counting its own training compute against California SB 53's 10^26 threshold is self-classification under a coverage scope threshold, not a designation by an authority.
A developer's internal decision that a model has crossed its own capability threshold is a threshold status, not an external designation.

Related terms

Related AI safety terms

Terms defined under this element that are not NIKOLAI elements themselves, each as the cited document uses it. Every term is also listed in the AI safety glossary.

Covered frontier model: Covered frontier model is the US Executive Order 14409 label for an AI model that the Director of NSA designates as crossing a threshold set by a classified benchmark of advanced cyber capabilities.; Source: Executive Order 14409
Frontier lab: Frontier lab, in Demis Hassabis's proposal, is an organisation with models that qualify as frontier-class on a standards body's benchmark thresholds; such labs would be encouraged to adopt best practices such as publishing model cards.; Source: Demis Hassabis, "A framework for frontier AI and the dawning of a new age"

Source of record

Where this definition comes from

EU AI Act Article 52 (Procedure), Article 52(4)
“"The Commission may designate a general-purpose AI model as presenting systemic risks, ex officio or following a qualified alert from the scientific panel pursuant to Article 90(1), point (a), on the basis of criteria set out in Annex XIII."”
https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-52
Executive Order 14409, Sec. 3(a)
“"develop and maintain a classified benchmarking process to assess the advanced cyber capabilities of AI models and determine the threshold at which an AI model should be designated a 'covered frontier model' for the purposes of this order"”
https://www.whitehouse.gov/presidential-actions/2026/06/promoting-advanced-artificial-intelligence-innovation-and-security/

Crosswalk

How named organisations use this concept

Every row below is a shadow mapping. A shadow row is CASRAI's own reading of a published document. No lab, evaluator or regulator named on a shadow row has declared, endorsed, or been consulted on it. That changes only when an organisation files its own Mapping Declaration.

Organisation	Their term, as published	Match & verification	Source
EUShadow mapping AI Act Articles 51-52	“Article 51(1)(b): classification "based on a decision of the Commission, ex officio or following a qualified alert from the scientific panel". Article 52(4): "The Commission may designate a general-purpose AI model as presenting systemic risks". Article 52(6): "The Commission shall ensure that a list of general-purpose AI models with systemic risk is published and shall keep that list up to date".” Article 52(5) lets a provider request reassessment, "at the earliest six months after the designation decision".	exactEQ confidence: high	EU AI Act Article 52 (Procedure)
US Government (Executive Order 14409)Shadow mapping Executive Order 14409	“A threshold "at which an AI model should be designated a 'covered frontier model' for the purposes of this order"; "Such a determination shall be made by the Director of NSA, in consultation with the National Cyber Director, the APST, the Director of CISA, and other representatives of the Department of War, as appropriate" (Sec. 3(a)).” Cyber capabilities only, and the benchmarking process is classified.	exactEQ confidence: high	Executive Order 14409
Demis Hassabis (personal essay)Shadow mapping "A framework for frontier AI" essay	“"A model would qualify as 'Frontier-class' if it meets certain thresholds on a set of benchmarks determined by the Standards Body and regularly updated to keep pace with evolving AI capabilities. Organisations with 'Frontier Models' as defined by those benchmarks would be deemed 'Frontier Labs'" (§Framework, ¶2).” Personal essay, not an official DeepMind policy document.	closeCL confidence: medium	Demis Hassabis, "A framework for frontier AI and the dawning of a new age"

What do these codes mean?

exact: The source term is equivalent to this element
close: The source term is close but not equivalent to this element
broad: The source term is broader than this element
narrow: The source term is narrower than this element
none: No mapping claim — used for false-friend and declared-but-undefined rows
EQ: Equivalent
CL: Close
BR: Source is broader than the element
NR: Source is narrower than the element
FF: False friend — same or similar label, different meaning
DU: Declared but undefined by the source
UV: Unverified

Related, not mapped

Pointers that are not crosswalk claims

These sources mention this concept but do not define or map it clearly enough to count as a crosswalk row — noted here so the research is visible without overstating it as a mapping.

California SB 53
No designation step: coverage follows from the developer's own compute count. SB 53 instead asks the Department of Technology to review the definitions yearly, weighing "The external verifiability of determining whether a person or foundation model is covered, with an aim toward definitions that are verifiable by parties other than the frontier developer" (22757.14(b)(5)).
California SB 53

Divergence

Where sources materially disagree

The three designation routes differ in who decides and what is public. The EU Commission designates on Annex XIII criteria and must publish a list of models with systemic risk; under Executive Order 14409 the Director of NSA designates against a classified cyber benchmark; Hassabis proposes a standards body using a published, regularly refreshed benchmark set. The EU route can override a developer's own view; the essay's route also extends the label from the model to its developer ("Frontier Labs").

Gap

Only one of the three routes (the EU) is in force with a public list requirement. The research did not find a published list of EU designations or of EO 14409 covered frontier models in the source corpus.