What Amazon, Meta, Uber, Anthropic and Others Want in the US AI Action Plan

As the White House finalizes its AI Action Plan, several tech giants, AI startups and even financial institutions are weighing in.

Amazon, Anthropic, Meta, Microsoft, Mistral AI, Uber, CrowdStrike and JPMorgan Chase are among the thousands that have voiced their views on how the U.S. should regulate artificial intelligence (AI) for growth and innovation. The federal government released all of the comments Thursday (April 24) in a searchable database.

OpenAI was one of the earlier submissions, calling for stronger export controls to China but less onerous ones for democratic countries, among other recommendations.

Google irked entertainment industry creators and other workers by calling for fair use of copyrighted content — echoed by OpenAI — along with recommendations for government AI adoption.

Overall, the submissions show several similarities — the need for infrastructure investment, open innovation and regulatory consistency to unify the patchwork of state laws — though they differ on approach.

Read more: OpenAI Calls for Federal Government to Preempt State AI Laws

Here’s a roundup of what tech giants, AI startups and other companies want from the Trump administration:

Amazon is pushing for energy infrastructure investments, cloud and chip access, workforce development, federal AI adoption and setting interoperable international standards.

• Streamline energy regulations: Amazon said AI will require a lot of electricity to power. It advocates for streamlining nuclear power projects and transmission upgrades to keep the U.S. competitive.
• Lead global talks in AI: Amazon urges the White House to lead in AI efforts globally by promoting global regulatory interoperability through international standards for AI adoption.
• Educate the workforce in AI: Americans need to be educated on practical AI implementation not just advanced technical training, Amazon says, and the U.S. must invest in more advanced AI researchers and engineers while also enabling workers to use AI tools at work.
• Transform government agencies using AI: Amazon says federal agencies should leverage AI and the cloud to move away from legacy, on-premises data centers to transform their operations.

Anthropic projects that as early as late 2026, frontier AI systems may rival Nobel laureates in reasoning ability — and must be treated as national assets. Here are three key items on its wish list:

• AI threat testing: Anthropic is calling for building robust federal infrastructure for testing powerful models for risks in cybersecurity and development of biological weapons, among others.
• Strengthen chip export controls: The company supports new restrictions on advanced chips — including the currently exempt Nvidia H20 — and agreements with other countries to prevent smuggling.
• Energy for AI: Echoing Amazon, Anthropic projects that 50 gigawatts of additional power will be needed by 2027 just for U.S. AI developers.
• Monitor economic impacts of AI: The U.S. government should enhance its data collection mechanisms to better capture the economic fallout from wider AI adoption, as well as prepare for “significant” changes.

Meta’s Llama models are at the heart of its vision for open-source AI leadership and its recommendations to the U.S. government reflect this priority.

• Don’t stifle open source: Meta calls on the U.S. to resist pressure to regulate or restrict open AI models, warning that doing so would empower authoritarian regimes to fill the vacuum.
• Federal agency adoption: Meta urges the U.S. government to use open models wherever possible, arguing they are more secure, customizable and better for national security use cases.
• Fair use clarity: It wants an executive order clarifying that training AI on public data is fair use, which is crucial against copyright lawsuits. This reflects the stances of OpenAI and Google.
• State rules harm innovation: Meta is warning that fragmented state-level rules will raise compliance costs and stifle innovation.

Microsoft said the U.S. must stay at the forefront of artificial intelligence, noting that it is investing more than $50 billion in U.S. AI infrastructure in 2025.

• Enhance computational and energy resources: AI needs a lot of power and the software giant called for modernizing the electric grid, permitting data center construction and enhancing U.S. manufacturing of critical grid components and AI hardware.
• Access to high-quality data: The company wants to unlock government and other publicly funded data for AI training.
• Promote trust, safety and national security with AI: Microsoft supports laws targeting deepfake fraud, harnessing AI in defense, and to advance cybersecurity protection.
• Upskill the U.S. workforce: The company says government should lead national efforts to educate about AI and prepare them for future jobs that will be driven by an AI economy.

France-based Mistral, with operations in Palo Alto, Calif., positioned itself as the startup champion for open-source innovation.

• Support open source: Like Meta, Mistral makes the case that transparency and public access to model weights improve research, security, and democratization of AI development.
• Weaken monopolies: Mistral advocates for antitrust enforcement, such as those brought by the Department of Justice (DOJ) and Federal Trade Commission (FTC), to ensure startups and small- to medium-size businesses (SMBs) can compete against incumbents.
• Enhance global chip trade: Mistral said overregulating chips or AI exports could inadvertently shift innovation to other countries.
• Global AI cooperation: The firm wants the U.S. to strike a balance — protecting national security while encouraging multinational innovation partnerships that can be mutually beneficial.

The ride-hailing company said in its remarks that AI plays an increasingly important role in shaping mobility services. Uber has invested in robust AI governance to ensure accountability in decisions driven by AI.

• Avoid overregulating low-risk AI: Many mobility-related AI applications pose minimal risk and should not be burdened with complex new rules, Uber says.
• Stop patchwork of state rules: Uber urges federal preemption to eliminate the growing patchwork of inconsistent state AI laws.
• Use existing laws first: Current regulations on privacy, discrimination and consumer protection already address most AI-related risks, Uber says.
• Adopt a risk-based framework: The company says regulations should focus on high-risk use cases and boost innovation in less risky ones such as pricing.

CrowdStrike’s comments center on using and securing AI in the cybersecurity domain. Last year, the cybersecurity company botched an update that cause a global outage of Windows computers and servers.

• Focus on AI for cybersecurity: CrowdStrike says that the use of AI to detect cyber threats gives the U.S. an “enormous” advantage because it can defeat new threats based on behavior rather than known signatures.
• Regulation should not stifle innovation: New regulations on AI should not harm innovation and development of new technologies, the company says.
• Protect the models: The company calls for robust protections around the AI systems themselves and training data for resilience.

JPMorgan, which runs more than 500 AI and machine learning systems, calls for greater AI governance.

• Use existing frameworks: The bank argued that current banking regulations — such as model and third-party risk management, data governance, and privacy — are already well-suited to handle AI.
• Sector-specific regulation: JPMorgan supports a sector-by-sector approach, where existing financial regulators, such as the OCC or Federal Reserve, lead AI oversight for banks.
• Level the playing field: The bank wants non-banks offering financial services to be subject to the same standards, especially for AI in credit underwriting and fraud detection.
• Unify federal and state regulation: JPMorgan echoes the concerns of others about a patchwork of state laws, and calls for the federal government to preempt state laws.

The post What Amazon, Meta, Uber, Anthropic and Others Want in the US AI Action Plan appeared first on PYMNTS.com.