Google also called for legalizing AI training on copyrighted content “without significant impact on copyright holders”

Google, following OpenAI, has released proposals in response to the Trump administration’s call for a national “AI Action Plan.” The tech giant advocates loosening copyright regulations around AI training and implementing “balanced” export controls that “protect national security while enabling U.S. exports and global business operations.”

Image source: Alex Dudar / unsplash.com

«“The United States needs to pursue a proactive international export policy to uphold American values ​​and support AI innovation globally. For too long, AI policy has focused disproportionately on threats, often ignoring the costs that ill-considered government interventions can have on innovation, national competitiveness, and scientific leadership—a dynamic that has begun to change under the new administration,” Google said.

For the development of AI systems and AI-related scientific innovation, “exceptions for fair use, text mining, and data mining are critical. These exceptions will allow the use of copyrighted public domain materials for AI training without significantly affecting copyright holders and avoiding the often highly unpredictable, unbalanced negotiations with data owners during model development or scientific experiments,” the company added.

Google is believed to have trained a number of models on copyrighted public data and is now fending off lawsuits from copyright holders who accuse the company of failing to notify them and pay compensation in advance. American courts have not yet taken a unified position on whether AI developers can use such data in good faith. Google has also expressed disagreement with the export control measures introduced under Joe Biden, which “impose a disproportionate burden on U.S. cloud service providers” and “may undermine economic competitiveness goals.” Notably, Microsoft, which operates in the same area, was not bothered by this aspect, and the corporation has previously expressed confidence that it will be able to “fully comply” with the sanctions.

Image Source: Mitchell Luo / unsplash.com

Google recalled that the US federal government has recently begun cutting spending and eliminating grants, and proposed instead to make “long-term, continuous” investments in fundamental research and development. The company also proposed that the government publish data sets that could be useful for commercial AI training, allocate funding for early-stage research and development, and make computing resources “widely available” to scientists and institutions.

Another important area is the legislative framework, which is still lacking in the US, despite the urgent need for comprehensive privacy and security standards. In the first two months of 2025 alone, the number of pending AI bills in the US has grown to 781. Google has warned authorities against imposing onerous obligations on AI developers, such as liability for the use of models. Last year, the company spoke out against the failed California regional bill SB 1047, which listed the precautions that creators of AI models must take before releasing them, as well as cases in which the creator can be held liable for harm caused by using the model.

The company believes that responsibility can be shifted to developers of end-user applications that rely on AI models created by others. Google has called disclosure requirements such as those in the European AI Act “overly broad,” and said the U.S. government should oppose regulations that “reveal trade secrets, allow competitors to duplicate products, or compromise national security by providing an adversary with a roadmap for how to circumvent defenses or simulate a hack.”