Get the weekly research letter
Companies change their terms quietly. We read every version and catch what actually changed. One email a week on the changes that matter and what they mean. No account.
The document states that Meta defines risk thresholds based on how much a model facilitates identified threat scenarios, and that the company applies mitigations to keep risks within what it characterizes as acceptable levels.
This analysis describes what Meta's agreement states, permits, or reserves. It does not constitute a legal determination about enforceability. Regulatory applicability and practical outcomes may vary by jurisdiction, enforcement context, and individual circumstances. Read our methodology
This provision describes the decision criteria Meta states it applies to model releases, but the specific thresholds, the definition of acceptable levels, and the mitigation mechanisms are not disclosed, limiting external assessment of the framework's operational rigor.
Interpretive note: The document references defined risk thresholds and acceptable levels but does not disclose the specific criteria, numerical standards, or governance structures used to establish or approve those thresholds.
The document states that risk thresholds inform model release decisions but does not disclose the specific criteria or numerical standards used to define acceptable risk levels, making independent evaluation of these thresholds not possible from this document alone.
Cross-platform context
See how other platforms handle Risk Threshold Establishment and similar clauses.
Compare across platforms →Monitoring
Meta has changed this document before.
Receive same-day alerts, structured change summaries, and monitoring for up to 25 platforms.
"Establishing risk thresholds: We define risk thresholds based on the extent to which our models facilitate the threat scenarios. We have processes in place to keep risks within acceptable levels, including applying mitigations.Excerpt from Meta's Frontier AI Framework
(1) REGULATORY LANDSCAPE: The EU AI Act requires high-risk AI providers to implement risk management systems with documented risk levels and residual risk assessments. The absence of disclosed threshold criteria in this document may be relevant to organizations assessing whether Meta's framework satisfies EU AI Act documentation standards. (2) GOVERNANCE EXPOSURE: Medium. The document asserts that acceptable risk levels are defined and maintained but does not disclose the standards used to establish those levels, the parties responsible for approving thresholds, or the conditions under which thresholds are revised. (3) JURISDICTION FLAGS: EU and EEA jurisdictions impose documentation and transparency obligations on AI providers that may require more granular threshold disclosure than this document provides. (4) CONTRACT AND VENDOR IMPLICATIONS: Organizations contracting with Meta for AI services should assess whether the undisclosed threshold framework is sufficient for their own internal AI governance documentation or regulatory obligations. (5) COMPLIANCE CONSIDERATIONS: Compliance teams should evaluate whether this framework's threshold description meets the specificity required by internal AI governance policies or applicable regulatory guidance, and whether supplementary disclosure requests to Meta are warranted.
This provision describes the decision criteria Meta states it applies to model releases, but the specific thresholds, the definition of acceptable levels, and the mitigation mechanisms are not disclosed, limiting external assessment of the framework's operational rigor.
The document states that risk thresholds inform model release decisions but does not disclose the specific criteria or numerical standards used to define acceptable risk levels, making independent evaluation of these thresholds not possible from this document alone.
No. ConductAtlas is an independent monitoring service. We are not affiliated with, endorsed by, or sponsored by Meta.