LIVE
EU AI Act enforcement begins · June 2026NIST AI RMF — risk management framework publishedISO/IEC 42001 AI management standard now certifiableOpenAI o3 sets new reasoning benchmarksAnthropic raises $4B Series EEU AI Act enforcement begins · June 2026NIST AI RMF — risk management framework publishedISO/IEC 42001 AI management standard now certifiableOpenAI o3 sets new reasoning benchmarksAnthropic raises $4B Series EEU AI Act enforcement begins · June 2026NIST AI RMF — risk management framework publishedISO/IEC 42001 AI management standard now certifiableOpenAI o3 sets new reasoning benchmarksAnthropic raises $4B Series E

US Executive Order on Safe AI (EO 14110)

Active
high risk
United States

NIST, CISA, DOD, DHS, OMB, Commerce Department, OSTP. Enforced through agency regulations, procurement requirements, and federal contractor obligations.

October 2023 - Active (some provisions under review by new administration, Jan 2025)

Official Text

Status

Active

Risk Level

High

Jurisdiction

United States

Enforcement

October 2023 - Active (some provisions under review by new administration, Jan 2025)

high risk framework

Primarily: US frontier AI developers (>10^26 FLOPS training). Secondary: Federal agencies. Voluntary best practice for all US organizations.

Overview

Executive Order 14110 on Safe, Secure, and Trustworthy AI, signed October 30, 2023. Requires safety testing reports from frontier AI developers, establishes federal AI governance, and directs 50+ agencies to develop sector-specific AI guidance. Key threshold: models trained with >10^26 FLOPs must report to government.

Scope

Directly mandatory for: (1) US frontier AI developers training models >10^26 FLOPs, (2) Federal agencies per OMB implementation guidance. Indirectly affects all AI companies seeking federal contracts or operating in regulated sectors.

Applicability

Who Is Affected

  • Frontier AI developers training models >10^26 FLOPs (OpenAI, Anthropic, Google, Meta, Microsoft)
  • All US federal agencies: must comply with OMB M-24-10 AI governance requirements
  • Federal contractors with AI in critical systems
  • Life sciences and biotech AI companies (biosecurity screening)
  • Cloud computing providers offering AI services to the federal government

Who Is Exempt

  • AI systems trained below the 10^26 FLOPs threshold (most commercial AI)
  • Non-US companies with no federal contracts (partially)
  • Open-source developers (some partial exemptions)

Key Prohibitions

  • Federal agencies cannot deploy AI in high-impact functions without completing required governance (per OMB)
  • Frontier AI companies must report capability evaluations - not doing so violates Defense Production Act obligations
  • AI in federal HR processes must not violate anti-discrimination laws
  • Deepfake generation for disinformation against US government: criminal liability

Key Requirements

  • Safety testing results for dual-use foundation models (>10^26 FLOPs training)
  • Red team evaluations shared with the US government (Defense Production Act authority)
  • NIST development of frontier AI safety guidelines and evaluation tools
  • Federal agencies to implement AI governance policies per OMB M-24-10
  • Screening for biological, chemical, nuclear, and radiological risks from AI
  • Watermarking standards for AI-generated content (NIST to develop)
  • Workforce impact assessments for federal AI deployment
  • Procurement: federal agencies prefer AI vendors demonstrating safety practices

Guardrails & Operational Controls

  • Pre-deployment safety testing: mandatory red-teaming for dual-use foundation models
  • Biosecurity: AI companies must implement screening to prevent models from assisting in weapons of mass destruction
  • Cybersecurity: AI models must not assist in cyberattacks on critical US infrastructure
  • Content provenance: NIST-led watermarking and labeling standards for AI-generated content
  • Privacy-preserving AI: federal agencies must minimize personal data in AI training

Implementation Guidance

  1. 1Determine if training compute exceeds 10^26 FLOPs threshold for mandatory reporting
  2. 2Prepare safety test results documentation for DHS/Commerce reporting requirements
  3. 3Implement red team evaluation processes before major model deployments
  4. 4Monitor NIST for new AI safety standards and evaluation frameworks under the EO
  5. 5Federal agencies: complete AI use case inventories and governance plans per OMB M-24-10

Industry Impact

Frontier AI Labs

Direct mandatory reporting obligations. Must share safety evaluations with US government.

critical

Federal Government

OMB M-24-10 mandatory compliance. Annual AI use case inventories. Human review for high-impact AI.

critical

Life Sciences / Biotech

NIAID and HHS developing biosecurity AI requirements. Screening for dual-use risks.

high

Defense Contractors

DoD AI Ethics Principles implementation required. New AI procurement requirements.

high

Cloud Providers

FedRAMP AI additions. Know-Your-Customer requirements for cloud compute services (IaaS).

high

Healthcare

HHS required to develop healthcare AI safety framework aligned with EO.

medium

Regulatory Timeline

PastCurrentUpcoming

Oct 30, 2023

EO 14110 signed by President Biden

Nov 2023

NIST AI Safety Institute (AISI) established at Commerce

Mar 2024

OMB M-24-10 issued - federal AI governance requirements

Jul 2024

NIST AI 600-1 (Generative AI profile) published

Jan 2025

EO 14110 revoked by Trump administration - some provisions persist via agency rules

2025

New administration AI policy developing - watch for replacement EO or legislation

Notable Enforcement Cases

  • 1All major frontier AI labs (OpenAI, Anthropic, Google, Meta, Microsoft, Amazon) voluntarily committed to safety practices under the EO framework
  • 2NIST established the AI Safety Institute within 30 days of EO signing - fastest agency implementation
  • 3EO revoked Jan 2025 by Trump administration; key agencies maintaining many provisions through existing authority

Penalties for Non-Compliance

Defense Production Act violations for non-reporting: regulatory and criminal exposure. Federal contractor debarment possible. Agency-specific fines for OMB non-compliance.

Framework Details

Short Name

US AI EO

Jurisdiction

United States

Enforcement Date

October 2023 - Active (some provisions under review by new administration, Jan 2025)

Enforcing Authority

NIST, CISA, DOD, DHS, OMB, Commerce Department, OSTP. Enforced through agency regulations, procurement requirements, and federal contractor obligations.

Status

Active

Risk Level

high

Affected Organizations

Primarily: US frontier AI developers (>10^26 FLOPS training). Secondary: Federal agencies. Voluntary best practice for all US organizations.

Exposure Areas

  • Frontier AI labs: mandatory reporting and red-team evaluation sharing with government
  • Federal agencies: non-compliance with OMB M-24-10 AI governance creates audit exposure
  • Life sciences AI: biosecurity screening obligation - misuse could trigger criminal liability
  • Government contractors: AI in federal contracts must demonstrate security and safety
  • Critical infrastructure AI: enhanced CISA oversight and vulnerability reporting

Tags

GovernmentFrontier AIEnterpriseUSSafety Testing

This is educational guidance only. Always consult qualified legal counsel for compliance decisions affecting your organization.