Google has laid out a broad AI policy agenda for the Trump administration, arguing that the U.S. should prioritize innovation, competitiveness, and scientific leadership as it develops a national “AI Action Plan.” The proposal puts the company firmly behind looser rules for AI training data, narrower burdens on cloud providers, and a federal approach to AI regulation.
The document also shows where Google sees the biggest policy risks for AI companies: copyright lawsuits, export controls, fragmented state laws, liability for downstream misuse, and transparency requirements that could expose sensitive technical information.
Google’s case for AI training on public data
One of the central points in Google’s proposal is copyright. The company argues that “fair use and text-and-data mining exceptions” are “critical” to AI development and AI-related scientific innovation.
In practical terms, Google wants policymakers to protect the ability of AI developers to train models on publicly available data, including copyrighted material, with limited restrictions. The company says these exceptions let developers use copyrighted, publicly available material for AI training without significantly affecting rightsholders.
Google also frames the issue as a speed and predictability problem. It says AI model development and scientific experimentation can be slowed by negotiations with data holders that are highly unpredictable, imbalanced, and lengthy.
This position comes as Google is fighting lawsuits from data owners who accuse the company of using public, copyrighted data without notice or compensation. The source article notes that U.S. courts have not yet decided whether fair use doctrine shields AI developers from intellectual property litigation in this context.
Export controls are another major target
Google also takes aim at certain export controls imposed under the Biden administration. The company argues that these rules “may undermine economic competitiveness goals” by placing disproportionate burdens on U.S. cloud service providers.
The rules are designed to limit access to advanced AI chips in disfavored countries. They also include exemptions for trusted businesses seeking large clusters of chips.
Google’s stance differs from at least one major competitor. Microsoft said in January that it was “confident” it could “comply fully” with the rules. Google, by contrast, is asking for what it calls “balanced” export controls that protect national security while still enabling U.S. exports and global business operations.
The broader policy argument is clear: Google wants the U.S. to remain active internationally while avoiding rules it believes could weaken American AI companies abroad. In the proposal, Google wrote that “The U.S. needs to pursue an active international economic policy to advocate for American values and support AI innovation internationally.”
Federal AI law over state-by-state rules
Google also warns that the U.S. regulatory environment is becoming chaotic because of a patchwork of state AI laws. The company urges the federal government to pass AI legislation, including a comprehensive privacy and security framework.
The source article says that just over two months into 2025, the number of pending AI bills in the U.S. has grown to 781, according to an online tracking tool. For a company operating across markets and product lines, that kind of fragmented rulemaking can create compliance complexity.
Google’s proposal does not only ask for fewer rules. It asks for rules at the federal level that would create a more consistent framework for AI developers, deployers, scientists, institutions, and commercial users.
The company also calls for “long-term, sustained” investments in foundational domestic R&D. It pushes back against recent federal efforts to reduce spending and eliminate grant awards, and says the government should release datasets that could help commercial AI training.
Google further recommends funding for “early-market R&D” and says computing and models should be “widely available” to scientists and institutions.
Liability and transparency remain flashpoints
Google is also asking the government to avoid obligations it sees as too heavy for AI system developers. One example is usage liability. The company argues that, in many cases, the developer of a model “has little to no visibility or control” over how that model is used.
That position reflects a broader industry debate over responsibility for AI harms. Google has historically opposed laws like California’s defeated SB 1047, which defined precautions an AI developer should take before releasing a model and described cases where developers might be held liable for harms caused by a model.
In Google’s view, deployers may often be better positioned than model developers to understand downstream risks, manage those risks, and monitor systems after release. The company wrote that “Even in cases where a developer provides a model directly to deployers, deployers will often be best placed to understand the risks of downstream uses, implement effective risk management, and conduct post-market monitoring and logging.”
Disclosure rules are another concern. Google calls some requirements being considered by the EU “overly broad” and says the U.S. government should oppose transparency mandates that require companies to reveal trade secrets, help competitors duplicate products, or compromise national security by giving adversaries a roadmap to bypass protections or jailbreak models.
The source article notes that a growing number of countries and states have passed laws requiring AI developers to disclose more about their systems. California’s AB 2013 requires companies developing AI systems to publish a high-level summary of the datasets used to train them. In the EU, once the AI Act comes into force, companies will have to provide model deployers with detailed instructions on the model’s operation, limitations, and risks.
What the proposal signals
Google’s AI policy proposal is not a narrow technical filing. It is a broad request for the U.S. government to shape AI rules around competitiveness, export capacity, research investment, and flexibility for developers.
The company is arguing that AI policy has focused too much on risk and not enough on the cost of regulation. As Google put it, “For too long, AI policymaking has paid disproportionate attention to the risks, often ignoring the costs that misguided regulation can have on innovation, national competitiveness, and scientific leadership — a dynamic that is beginning to shift under the new Administration.”
The stakes are substantial for AI companies. Copyright rules could determine what data can be used for training. Export controls could shape where advanced AI infrastructure is available. Federal legislation could either simplify or complicate compliance. Liability and transparency rules could influence how AI models are released, documented, and monitored.
Google’s message is that the U.S. should move toward a lighter and more unified AI policy framework. Whether courts, lawmakers, regulators, and international partners accept that approach remains unresolved in the source article.