{"slug":"frontier-model-safety-framework","id":"AIPT-CTL-0014","title":"Frontier model safety and security framework","kind":"policy","kind_label":"Policy","purpose":"Commits a developer of the most capable general-purpose models to a published, regularly reviewed description of how it identifies catastrophic and systemic risks, decides when a model is safe to train further or release, and protects model weights.\n","description":"The developer writes and publishes a framework that sets capability thresholds of concern, the evaluations used to detect them, the mitigations that must be in place before crossing each threshold, the security measures for weights and infrastructure, and the internal governance that makes go or no-go decisions. Training compute is tracked against regulatory thresholds so that notification duties are met on time. The framework is reviewed at least annually and after significant capability jumps or incidents, changes are recorded in a version log, and summaries of evaluations and mitigations are shared with authorities and downstream providers as required.\n","owner_role":"Head of AI safety","frequency":"annual","frequency_label":"Annual","risk_subdomains":["7.2","7.1","4.2","6.1"],"evidence":[{"type":"policy_document","title":"Published frontier safety framework","description":"Capability thresholds, evaluations, mitigations, security measures and governance, with a version log."},{"type":"evaluation_report","title":"Dangerous-capability evaluation report","description":null},{"type":"regulatory_filing","title":"Threshold notification to an authority","description":null},{"type":"register_entry","title":"Training compute tracking record","description":null}],"framework_references":[{"framework":"nist_ai_rmf","framework_name":"NIST AI RMF 1.0","reference":"GOVERN 1.3, 1.4; MAP 5.1; MEASURE 2.6; MANAGE 1.3","note":null,"confidence_level":"medium"},{"framework":"iso_42001","framework_name":"ISO/IEC 42001:2023","reference":"Clause 5.2, 6.1.2; Annex A.6.1.2","note":null,"confidence_level":"low"},{"framework":"mitre_atlas","framework_name":"MITRE ATLAS","reference":"AML.M0001 Limit Model Artifact Release, AML.M0005 Control Access to ML Models and Data at Rest","note":null,"confidence_level":"medium"}],"obligations":[{"slug":"eu-ai-act-gpai-systemic-risk","title":"Manage systemic risk for high-impact general-purpose models","relationship":"satisfies","note":"Compute tracking, notification and the safety and security framework.","confidence_level":"medium","is_binding":true,"policy":"eu-ai-act","jurisdiction":"eu","url":"https://aipolicytracker.org/obligations/eu-ai-act-gpai-systemic-risk"},{"slug":"eu-ai-act-art-52-systemic-risk-notification","title":"Providers of GPAI models must notify the Commission within two weeks of meeting the systemic-risk threshold","relationship":"satisfies","note":"Compute tracking and threshold notification are part of the framework.","confidence_level":"high","is_binding":true,"policy":"eu-ai-act","jurisdiction":"eu","url":"https://aipolicytracker.org/obligations/eu-ai-act-art-52-systemic-risk-notification"},{"slug":"eu-ai-act-art-55-systemic-risk-incident-reporting","title":"Providers of systemic-risk GPAI models must track and report serious incidents to the AI Office","relationship":"supports","note":"Framework defines the incident types and escalation.","confidence_level":"medium","is_binding":true,"policy":"eu-ai-act","jurisdiction":"eu","url":"https://aipolicytracker.org/obligations/eu-ai-act-art-55-systemic-risk-incident-reporting"},{"slug":"us-california-sb-53-frontier-ai-framework","title":"Large frontier developers must publish a frontier AI framework","relationship":"satisfies","note":"The published, annually reviewed framework with a change log.","confidence_level":"high","is_binding":true,"policy":"us-california-sb-53","jurisdiction":"us-california","url":"https://aipolicytracker.org/obligations/us-california-sb-53-frontier-ai-framework"},{"slug":"us-california-sb-53-critical-safety-incident-reporting","title":"Report critical safety incidents to the Office of Emergency Services","relationship":"supports","note":"Defines the incident types the framework is meant to prevent.","confidence_level":"medium","is_binding":true,"policy":"us-california-sb-53","jurisdiction":"us-california","url":"https://aipolicytracker.org/obligations/us-california-sb-53-critical-safety-incident-reporting"},{"slug":"us-california-sb-53-transparency-report","title":"Frontier developers must publish a transparency report before deploying a new frontier model","relationship":"supports","note":"Report describes steps taken under the framework.","confidence_level":"medium","is_binding":true,"policy":"us-california-sb-53","jurisdiction":"us-california","url":"https://aipolicytracker.org/obligations/us-california-sb-53-transparency-report"},{"slug":"us-california-sb-53-catastrophic-risk-assessment-summaries","title":"Large frontier developers must send periodic summaries of catastrophic-risk assessments to the state","relationship":"satisfies","note":"Framework defines the catastrophic-risk assessment whose summaries are submitted.","confidence_level":"medium","is_binding":true,"policy":"us-california-sb-53","jurisdiction":"us-california","url":"https://aipolicytracker.org/obligations/us-california-sb-53-catastrophic-risk-assessment-summaries"},{"slug":"eu-ai-act-art-55-systemic-risk-cybersecurity","title":"Providers of systemic-risk GPAI models must secure the model and its infrastructure","relationship":"supports","note":"Framework commits to the weight-security measures.","confidence_level":"high","is_binding":true,"policy":"eu-ai-act","jurisdiction":"eu","url":"https://aipolicytracker.org/obligations/eu-ai-act-art-55-systemic-risk-cybersecurity"},{"slug":"south-korea-ai-basic-act-art-32-safety-measures-for-high-performance-ai","title":"Operators of AI above the compute threshold must run lifecycle risk management and report safety results","relationship":"satisfies","note":"Compute tracking, lifecycle risk assessment and reporting to the authority.","confidence_level":"medium","is_binding":true,"policy":"south-korea-framework-act-on-the-development-of-artificial-intelligence-and-establishment-of-a-foundation-for","jurisdiction":"south-korea","url":"https://aipolicytracker.org/obligations/south-korea-ai-basic-act-art-32-safety-measures-for-high-performance-ai"}],"related_controls":["genai-red-team-testing","ai-risk-assessment","ai-incident-management-and-reporting"],"review_status":"verified","confidence_level":"medium","last_verified_at":"2026-09-26","reviewed_by":"Bhaskar Bhatt","updated_at":"2026-10-15T06:46:22+00:00","url":"https://aipolicytracker.org/controls/frontier-model-safety-framework","context_url":"https://aipolicytracker.org/controls/frontier-model-safety-framework.md"}