Photo Credit: BoliviaInteligente
Google publishes a paper outlining its proposed two-pronged approach to AI governance in the United States, doubling down on ‘fair use’ AI training.
Late last week, Google published a policy paper outlining a proposed approach to AI regulation in the United States that separates oversight of widely deployed AI applications from pioneering AI models. To that end, the company calls for a “pragmatic, evidence-based framework” that treats advanced systems differently than tools like chatbots.
Unsurprisingly, the paper argues that AI policy should focus on the output of a given model rather than the inputs used to create it. Google proposes the creation of a Frontier AI Regulatory Organization (FARO), an industry-funded body that would operate under federal oversight and develop standards for security, safety, transparency, and incident reporting.
According to Google, FARO could set scientific benchmarks for the capabilities of frontier models in areas like cybersecurity, as well as chemical, biological, radiological, and nuclear risks. FARO could also oversee independent audits and require frontier AI companies to publish and follow safety frameworks before releasing highly advanced models.
On the other side of the aisle, Google argues that the federal government should primarily rely on existing legal frameworks when addressing widely deployed AI applications, with targeted updates as needed. Google says that policy should focus on “real-world outputs and specific harms” rather than “micromanaging the science” behind these tools.
“We believe a balanced approach can protect the public without paralyzing progress, ensuring everyone can benefit from AI,” wrote Kent Walker, President of Global Affairs at Google. “We hope these ideas contribute to the important ongoing dialogue, and we invite you to read the full paper.”
Google points out several priority areas, including child safety, information integrity, workforce preparedness, copyright, privacy, and energy infrastructure for data centers. The company also supports measures like AI interaction guidelines for children, disclosures that chatbots are not sentient, rules for self-harm related queries, watermarking and provenance standards for generative AI, workforce reskilling, and privacy-enhancing technologies.