Google, Follow Openai’s heels,,,,, Policy proposals were issued In response to the Trump administration’s call for the formulation of a national “AI Action Plan.” The tech giant acknowledged weak copyright restrictions on AI training and “balanced” export control, “protecting national security while enabling U.S. exports and global business operations.”
“The United States needs to adopt positive international economic policies to advocate American values and support AI innovation internationally,” Google wrote in the document. “For a long time, AI policy formulation has focused too disproportionately on risks, often overlooking the costs of misleading regulation in innovation, national competitiveness and scientific leadership, a dynamic that is beginning to shift under a new government.”
One of Google’s more controversial suggestions has to do with the use of IP-protected materials.
Google believes that “rational use and text and data mining exceptions” are crucial for AI development and scientific innovations related to AI. Like Openaithe company aims to write it and its competitors as the right to publicly available data, including copyrighted data.
“These exceptions allow AI training using copyrighted, publicly available materials without significantly affecting the right shell and avoid highly unpredictable, imbalanced and lengthy negotiations with data holders during model development or scientific experiments,” Google wrote.
Google, there It is said that Many publicly trained models, copyrighted data are fighting litigation Data owners with allegations that the company failed to notify and compensate before doing so. The U.S. court has not yet determined whether effective use of doctrine can effectively protect AI developers from IP lawsuits.
In its proposal, Google also encountered problems with export controls implemented by Biden Administration, which said it “could undermine economic competition goals” and “put a disproportionate burden on U.S. cloud service providers.” This is in stark contrast to statements from Google competitors such as Microsoft, who said in January that it can be “confidence” and that it can fully comply with the rules.
Importantly, export rules provide exemptions for trusted companies seeking a large number of advanced AI chip groups.
Elsewhere in the proposal, Google calls for a “long-term, ongoing” investment in basic domestic research and development, which opposes recent federal efforts to reduce spending and cancel grant awards. The company said the government should release data sets that could help with commercial AI training and allocate funds to “early market research and development” while ensuring computing and models are “widely available” to scientists and institutions.
Google pointed to a chaotic regulatory environment pieced together by U.S. national AI laws, urging the government to pass federal legislation for AI, including a unified privacy and security framework. It’s only two months until 2025 The number of AI bills pending in the United States has grown to 781according to online tracking tools.
Google warns the U.S. government not to impose its heavy obligations to AI systems, such as the obligation to use liability. Google believes that in many cases, developers of models have “few visibility or control over how developers use models” and should not be held liable for abuse.
“Even if developers provide models directly to deployers, deployers are often best suited to understand the risks of downstream use, implement effective risk management, and conduct aftermarketing.
Monitor and record,” Google wrote.
Google also called the disclosure requirements considered for the EU “too broad” and said the U.S. government should oppose the need to “learn trade secrets, allow competitors to copy products or compromise national security rules by providing adversaries with a roadmap on how to bypass adversaries who support protection or jailbreak models.”
