AI Compliance for Frontier Models
Frontier Models is addressed by 50 regulatory updates across 8 jurisdictions and 6 frameworks. This page tracks how regulators worldwide are approaching frontier models in the context of artificial intelligence.
Framework Requirements for Frontier Models
Regulations Covering Frontier Models
European Union(26)
AI Act Guidelines Implementation Update
The Commission says it is preparing additional AI Act guidelines, including on high-risk systems, the AI value chain, substantial modification, post-market monitoring and interplay with other EU laws.
EU AI Act: draft implementing act for AI regulatory sandboxes
The Commission opened feedback on a draft implementing act that would set common rules for establishing and operating AI regulatory sandboxes under the AI Act.
EU AI Act: draft guidelines on classification of high-risk AI systems
The Commission opened a consultation on draft guidelines to help providers, deployers, and market surveillance authorities assess whether an AI system is high-risk.
European AI Office and AI Act governance framework
The European AI Office is established within the European Commission to support, enforce, and coordinate AI Act governance, especially for general-purpose AI.
ESAs warn on external dependencies, cyber threats and private credit risks
The ESAs identify external dependencies, cyber threats linked to capable AI models, and private credit as key vulnerabilities for the EU financial system and call for stronger preparedness and monitoring.
International(9)
NIST AI RMF revision and critical-infrastructure profile development
NIST says the AI RMF is being revised and that a new trustworthy-use profile for critical infrastructure is under development, while the baseline AI RMF 1.0 remains voluntary and headed for review no later than 2028.
NIST signals revision of AI RMF and new critical-infrastructure profile work
NIST’s AI RMF materials state that the framework is being revised and that a new critical-infrastructure trustworthy-AI profile was launched on 2026-04-07, so organizations relying on the RMF should track the revision now for shifting implementation guidance.
NIST SP 1353 AI-enabled CSF analysis and reporting draft
NIST issued the initial public draft of SP 1353 on 2026-08-19 and set comments due by 2026-10-15, so teams using AI for CSF analysis should assess the draft and submit feedback before the close date.
NIST draft misuse-risk guidance targets dual-use foundation models
NIST’s draft guidance on dual-use foundation models may affect model governance and testing workflows, so teams should treat it as an emerging reference for misuse-risk management even though it is not itself an AI RMF update.
NIST generative AI profile remains a key AI RMF companion resource
NIST’s July 26, 2024 Generative AI Profile remains a key companion to AI RMF 1.0, so organizations using GenAI should continue to map controls to the profile’s risk scenarios and governance themes.
US Federal(6)
FTC AI enforcement posture remains active across deceptive claims and AI investments
The FTC’s AI hub and related matters, including Rytr, Workado, and DoNotPay, show that the agency continues to police deceptive AI claims and conduct, so AI product and marketing teams should assume ongoing enforcement scrutiny now.
NIST AI RMF update process tied to White House AI Action Plan
NIST says the AI RMF 1.0 is being revised under the July 23, 2025 White House AI Action Plan, which means organizations relying on the framework should expect updated implementation guidance and profile changes rather than a static reference.
US voluntary AI commitments and frontier AI policy direction
White House voluntary AI commitments remain a benchmark for frontier AI governance, so firms should continue aligning safety testing, provenance, cybersecurity, and information-sharing controls even though the commitments are nonbinding.
FTC seeks public comment on AI accuracy policy statement
On 2026-07-07, the FTC proposed a policy statement on AI accuracy, signaling that claims and system behavior that manipulate expected accuracy may be treated as deceptive under Section 5.
Congressional AI-related bills remain active but are not a single federal AI regime
Several AI-adjacent bills are active in Congress, but they do not create a unified federal AI law, so federal compliance planning remains agency-driven rather than statute-driven.
Australia(4)
eSafety publishes transparency notice responses for AI services
eSafety publishes a transparency-notice report on AI services and explains how it uses enforceable notices to assess compliance with the Basic Online Safety Expectations.
Australia Age-Restricted Material Codes fully commence
Australia's Age-Restricted Material Codes have fully commenced, requiring online services across the ecosystem to put in place age-appropriate protections for children and age assurance for access to certain adult material.
Generative AI – position statement
eSafety’s generative AI position statement describes the technology, identifies risks and harms, and encourages Safety by Design, transparency, and accountability measures for services that generate content.
Australian Search Code registered with new AI protections
eSafety will register the Search Code with added protections to reduce the risk that search services return child abuse material or use integrated AI to generate synthetic versions of it.
United Kingdom(2)
FCA, Bank of England and Treasury joint statement on frontier AI models and cyber resilience
The FCA, Bank of England, and HM Treasury said firms must be able to identify, monitor, and manage external AI-related applications, libraries, and services integrated into their networks, raising the bar for cyber and third-party resilience.
FCA Mills Review on how AI will reshape retail financial services
The FCA launched a review of advanced AI’s impact on retail financial services, with feedback due 24 February 2026 and recommendations expected for the FCA Board in summer 2026.
Canada(1)
Singapore(1)
New York(1)
Weekly digest
Leave your email to get each issue in your inbox. Free, no account required.
We use your email only for the digest. Privacy policy