Part of: this research — view all 7 rows
As of November 14, 2026, no jurisdiction holds the authority required to resolve this question. The barrier is not a lack of regulatory appetite, but the rigid structural criteria of the mandate: the evaluations must be government-commissioned (rather than developer-paid), cover already-deployed models, and occur on a fixed recurring schedule. This combination runs directly counter to standard regulatory design. To conserve state capacity and budgets, legislatures overwhelmingly favor developer-funded compliance, self-reporting, or discretionary powers that can be deployed case-by-case.
Current and pending frameworks uniformly miss this strict standard. In the EU, the AI Act’s Articles 91-93 grant binding post-deployment evaluation powers, but the recent Commission Implementing Regulation (EU) 2026/1755 confirms these procedures are case-triggered, lacking a scheduled cycle EU AI Model Evaluation and Enforcement Rules Evaluations and Proceedings Implementing Act. In the US, state laws like California's SB 53 and New York's RAISE Act AI Safety: Governor Hochul Announces Next Steps to ... rely on self-reporting. Where independent audits are mandated, as in Illinois' SB 315 After the Midterms: Preparing for Frontier-Model Audits and ... and California's SB 813/AB 1405 California Establishes First-in-the-Nation AI Auditor ..., the developers retain and pay the auditors themselves. At the federal level, the pending FRONTIER Act relies on developer-retained verifiers and ad hoc assessments H.R.9925 - 119th Congress (2025-2026): FRONTIER Act, and any near-term mandates face strong headwinds from a deregulatory executive posture expected to last through at least early 2029 Promoting Advanced Artificial Intelligence Innovation and Security Democrats Plan AI and Data-Centre Bills for 2027.
The earliest plausible pathways, driving the 10th percentile to May 15, 2028, depend on ongoing legislative efforts adopting a stricter design. Australia is currently the strongest candidate following its November 10, 2026 announcement of a new AI safety regulator that will mandate "independent evaluation" of models https://www.pm.gov.au/media/new-australian-ai-safety-and-security-protections New Australian AI safety and security protections. If 2027 Australian legislation includes a fixed schedule for government-run tests Australia's AI Safety Regulator and Mandatory Incident ..., it could trigger resolution by 2028. Alternatively, California might convert recent executive recommendations (EO N-9-26) Governor Newsom issues executive order to accelerate ... California AI Executive Order Quickens Auditing Process into a 2027 bill that establishes state-commissioned, scheduled audits, provided the state itself dictates and funds the evaluators. Allowing for typical legislative and institutional implementation lags, November 15, 2029 is a more credible 25th-percentile date than assuming these early proposals quickly produce a qualifying regime.
Despite these near-term possibilities, a median resolution on April 15, 2032 is most credible. Comparison with related questions reinforces the need to allow for typical legislative and institutional implementation lags, even where political support emerges relatively quickly . A paradigm shift toward recurring, government-run testing will likely require a significant catalyst—such as a major post-deployment incident or sustained, rapid capability gains—to overcome the entrenched preference for cheaper, flexible compliance models. The 75th percentile falls on September 15, 2036, reflecting the possibility that both this political shift and the establishment of a qualifying institutional framework take substantially longer. The long right tail, extending to July 1, 2043 at the 90th percentile, accounts for the substantial probability that regulators permanently settle on update-triggered testing, discretionary investigations, and developer-funded audits, leaving this exact legislative structure unrealized for well over a decade.
Set against related questions, this distribution was shifted slightly later, moving the median from early 2031 to mid-2032, to better reflect typical legislative and institutional implementation lags .
Ask a followup
Sign in to run · $20 free credit, no card · every claim cited