Draft AI-specific Legislation

Proposed framework for AI regulation including development fund with 2% tech levy and explicit bans on autonomous lethal weapons systems development.

Analysis

The proposed AI regulation framework includes several key requirements:

2% technology levy on large tech companies to fund AI development and oversight. This levy would apply to companies with global revenue over $1 billion.

Explicit ban on autonomous lethal weapons systems development, in line with the UN Convention on Certain Conventional Weapons.

Mandatory AI impact assessments for high-risk AI systems, similar to those required under the EU AI Act.

Transparency requirements for AI systems, including clear disclosure when AI is being used to make decisions affecting individuals, as outlined in the OECD AI Principles.

Algorithmic auditing to detect and mitigate bias, discrimination and other harms, drawing on frameworks like the NIST AI Risk Management Framework.

Organizations face several challenges in complying with the proposed AI regulations:

Lack of standardized assessment methodologies for AI impact and risk, as highlighted in a Brookings Institution report.

Difficulty explaining complex AI models, especially deep learning systems, as noted in an MIT Technology Review article.

Balancing innovation with regulation, a challenge discussed in a World Economic Forum white paper.

Ensuring global interoperability of AI governance frameworks, as analyzed in an OECD report.

To effectively implement the proposed AI regulations, organizations should:

Establish a cross-functional AI governance team to oversee compliance efforts, as recommended in Deloitte's AI governance framework.

Implement robust documentation practices for AI development and deployment, following guidelines like those in the IEEE Ethically Aligned Design.

Conduct regular AI ethics training for employees, drawing on resources like the World Economic Forum's AI Ethics Toolkit.

Utilize AI governance platforms like IBM's AI Fairness 360 or Google's Responsible AI Toolkit to support compliance efforts.

As this is a draft proposal, there are no recent updates. However, related developments include:

The EU AI Act reaching political agreement in December 2023, with final adoption expected in 2024.

The proposed framework intersects with several existing regulations:

GDPR: Overlaps on data protection and privacy requirements for AI systems processing personal data.

EU Digital Services Act: Shares focus on algorithmic transparency and accountability.

NIST AI Risk Management Framework: Provides complementary guidance on AI risk assessment and mitigation.

The proposed regulations would significantly impact various industries:

Technology sector: Facing increased costs due to the 2% levy and compliance requirements, as analyzed in a McKinsey report.

Healthcare: Potentially slowing AI adoption due to stricter oversight, but improving patient trust, as discussed in a Nature Medicine article.

Financial services: Enhancing algorithmic fairness in lending and investment decisions, as explored in a World Economic Forum white paper.

2026-03-11 — Commerce Department must evaluate burdensome state AI laws by March 11, 2026, flagging those compelling disclosures or altering outputs in violation of constitutional rights, for referral to the Task Force.[1] (source)
2026-01-10 — DOJ AI Litigation Task Force begins operations on January 10, 2026, per Executive Order, to litigate against state AI laws on preemption and commerce clause grounds, signaling industry-wide legal battles.[3] (source)
2026-01-01 — California's TFAIA requires frontier AI developers to publish risk mitigation frameworks for catastrophic risks, alongside AB 2023 for generative AI training data disclosure and SB 942 for AI content detection, all effective January 1, 2026, amid federal preemption threats.[1] (source)
2025-12-11 — President Trump issued an Executive Order on December 11, 2025, establishing a federal AI policy framework that preempts inconsistent state AI laws, targeting California's TFAIA and Texas RAIGA effective January 1, 2026, while creating a DOJ AI Litigation Task Force.[1] (source)