US AI Action Plan and What it Means for Australia

AI

March 30, 2025

The Trump administration’s call for input on the development of a new US Artificial Intelligence (AI) Action Plan has prompted responses from leading AI companies: OpenAI, Google, and Anthropic. Aiming to enhance America’s AI dominance, these submissions reveal shared objectives in accelerating AI innovation and deployment, particularly against China. However, their policy recommendations differ. OpenAI prioritises federal regulatory streamlining, aggressive export controls on China, and broad fair use for AI training . Google also seeks copyright exceptions for training data and balanced export controls, while advocating for sustained government investment in foundational AI research . Anthropic, focused on AI safety, emphasises national security testing, tightened export controls, and enhanced security for AI labs . These proposals have implications for the Australian AI market, requiring stakeholders to understand the evolving international AI policy environment.

Introduction:

In early 2025, the Trump administration initiated the development of a new US AI Action Plan through Executive Order 14179, aiming to solidify America’s global AI leadership . This followed the revocation of the Biden administration’s AI policies . A public consultation invited input on priority policy actions . OpenAI, Google, and Anthropic responded with detailed proposals, which this report analyses for their core tenets, validity based on provided research, and potential impact, particularly on the Australian AI market. The rapid advancement of AI and increasing geopolitical competition underscore the urgency of these policy discussions.

OpenAI’s Policy Recommendations:

OpenAI’s submission advocated for “freedom of intelligence,” encompassing several key areas.

Regulatory Strategy and Innovation: OpenAI requested a “regulatory strategy that ensures the freedom to innovate,” focusing on voluntary partnerships with the federal government to avoid bureaucratic obstacles and prevent advantages for competitors like China. They called for federal legislation to override state laws, arguing that a fragmented regulatory landscape would hinder AI development.

Export Controls: OpenAI proposed promoting the global adoption of American AI while restricting access for adversaries, especially China. They suggested banning Chinese AI like DeepSeek, citing security risks due to its ties to the CCP. This stance reflects the geopolitical dimension of AI policy.

Copyright Strategy: OpenAI advocated for a copyright strategy that “promotes the freedom to learn,” enabling AI models to use copyrighted materials responsibly. They argued this is crucial for maintaining US technological progress against countries with less stringent copyright enforcement.

Infrastructure: OpenAI emphasised the need for significant government investment in AI infrastructure, including a massive power transmission buildout and open-sourcing government datasets.

Overarching Strategy: OpenAI’s overall view emphasised an active international economic policy to promote American values and AI innovation globally, arguing against overemphasis on risks in policymaking.

Government Adoption: OpenAI stressed the importance of government agencies adopting advanced AI tools to enhance efficiency and national security.

Google’s Policy Recommendations:

Google’s submission echoed some of OpenAI’s concerns while presenting distinct perspectives on research funding and international collaboration.

Copyright and Data Access: Similar to OpenAI, Google called for a fair use exemption for training AI data, arguing it could be allowed without significantly impacting rights holders. They highlighted the commercial challenges of negotiating with data rights holders.
Export Controls and Market Access: Google advocated for a revision of export controls to support market access for US businesses while targeting pertinent risks. They cautioned against burdens that could disadvantage US companies globally.
Government AI Adoption and Infrastructure: Google also supported AI adoption in government, uniform federal laws, and infrastructure spending.
Foundational R&D Funding: Unlike OpenAI’s focus on deregulation, Google highlighted the importance of sustained government investment in foundational AI research to maintain US leadership.
Safety and Liability: Google called for clear liability delineation between model providers and users and criticised EU safety disclosure requirements as overly broad, citing concerns about trade secrets and national security.
Pro-Innovation International Approaches: Google advocated for an active international economic policy, championing market-driven technical standards and working with allies to address national security risks, while also seeking to dismantle foreign AI barriers.

Anthropic’s Policy Recommendations:

Anthropic’s submission heavily emphasised AI safety and national security, aligning with their core mission.

National Security Testing: Anthropic stressed the imminence of AGI and recommended establishing national security testing for domestic and foreign AI models.
Strengthening Export Controls: Anthropic proposed significantly ramping up export controls, including government-to-government agreements for countries hosting large chip deployments.
Enhancing Lab Security: Anthropic recommended integrating AI labs with the intelligence community, including classified communication channels and expedited security clearances.
Scaling Energy Infrastructure: Similar to OpenAI and Google, Anthropic called for scaling energy infrastructure, proposing a target of 50 gigawatts of additional power by 2027 .
Accelerating Government AI Adoption: Anthropic also supported accelerating AI adoption within government to modernise operations.
Preparing for Economic Impacts: Uniquely, Anthropic emphasised preparing for the economic impact of AI, suggesting modernisation of economic data collection.
Concerns about Chinese Espionage: Anthropic CEO Dario Amodei expressed concerns about Chinese industrial espionage targeting American AI companies.

Comparative Analysis:

The proposals show consensus on infrastructure and government adoption but diverge on export controls and copyright.

Table 1: Comparison of Key Policy Recommendations

Category	OpenAI	Google	Anthropic
Regulation	Federal preemption	Uniform federal laws	National security frameworks
Export Controls	Ban Chinese AI , restrict non-allies , promote “democratic AI”	Balanced controls , support market access, target risks	Ramp up controls, government agreements for chips
Copyright	Broad fair use	Fair use exceptions , text/data mining	Not mentioned
Infrastructure	Massive buildout	Government spending , address energy	Scale energy (50 GW)
Government Adoption	Public-private partnerships	Lead by example	Government-wide review
Safety/Security	National competitiveness	Liability delineation , transparency concerns	National security testing , lab security, economic impact

Differing export control stances likely reflect strategic priorities, with OpenAI and Anthropic favouring a stronger stance against China, while Google balances security with market access. The shared interest in government adoption and infrastructure indicates mutual benefits and fundamental needs for AI ecosystem growth.

Implications for the US AI Action Plan:

The Trump administration’s focus on deregulation and American AI dominance may favor OpenAI’s call for federal preemption and the stricter export control proposals from OpenAI and Anthropic, aligning with national security concerns . Google’s emphasis on sustained R&D funding might face more scrutiny under a deregulatory approach . Infrastructure investment, a shared priority, could garner broader support . Copyright exceptions and the extent of export controls are potential areas of debate . The AI Action Plan is expected by July 2025.

Connecting to the Australian Market:

The Australian AI market is growing significantly, reaching an estimated $2.9 billion AUD in 2024 and projected to reach USD 9.81 billion by 2030 . Key growth areas include cybersecurity and medical imaging . The Australian government is actively developing AI policies, including ethics principles, a policy for responsible use in government , the National AI Capability Plan , and proposed mandatory AI guardrails .

US AI policy developments, particularly the AI Action Plan, could significantly impact Australia. Stricter US export controls might limit Australia’s access to advanced AI hardware, affecting its AI sector’s growth. The US approach to copyright for AI training could influence Australia’s policy considerations. Australia’s focus on ethical AI development aligns with potential US emphasis on safety, creating collaboration opportunities. Increased US government investment in AI and adoption could signal a global trend, benefiting Australian companies in related sectors. The US stance on Chinese AI might prompt Australia to assess its own reliance on international AI technologies. Australia’s National AI Capability Plan aims to boost investment, strengthen capabilities, enhance skills, and secure economic resilience in the AI sector. The government has also released proposed mandatory guardrails for high-risk AI settings and a Voluntary AI Safety Standard. Australia’s Policy for the responsible use of AI in government mandates accountability and transparency for AI adoption within public sector entities.

Conclusion:

The US AI Action Plan proposals from OpenAI, Google, and Anthropic reflect a shared goal of accelerating AI innovation and securing American leadership, albeit with differing priorities. The Trump administration’s approach will likely shape the final plan, potentially favouring deregulation and a strong stance against China. These US policy shifts have significant implications for the Australian AI market, affecting access to technology, influencing policy considerations, and potentially creating opportunities for collaboration. Australia’s AI ecosystem must remain informed and adaptable to navigate the evolving global AI landscape effectively.