US to draft comprehensive AI legislation

US Senator Chuck Schumer has proposed, at the legislative level, to conduct independent public audits of commercial AI products before their launch. The The Register reports.

The Senate majority leader said he had developed a framework for regulating the development, deployment, and use of advanced machine-learning technologies. He said the boom in generative AI and China’s actions to curb it have made the question of model oversight in the United States timely.

Schumer hopes that his initiative will lead to ‘comprehensive’ legislation that organisations will follow as they deploy AI.

“The era of AI has arrived, and it will linger here. It is time to develop, deploy and expand its potential to benefit our country for generations to come,” he said.

According to Schumer, leading industry experts and opinion leaders contributed to developing the AI regulation structure. He is convinced that the rules will prevent ‘potentially catastrophic damage’ to the United States while ensuring the country’s leadership in this field.

“But there is still much work ahead, and we must move quickly. I look forward to working across all fronts, […] to shape this proposal and refine the legislation,” Schumer said.

The senator has not yet published his plan. He says it is based on four principles: who, where, how, and protection.

Under the proposals, it must be clear who trained the model and for whom, where developers obtained the data, and how the technology will be used. This information should help independent researchers audit the algorithms before being released to the market, Schumer says.

In terms of safeguards, developers will have to demonstrate that their systems align with American values, and ‘AI developers fulfill the promise of creating a better world’.

Schumer promised to share the concept with leaders from academia, industry, and government in the near future. He also intends to work with think tanks and research institutes to refine the proposals.

He plans to turn the resulting policy into a bill.

In April, the Cyberspace Administration of China unveiled a draft of rules regulating generative AI algorithms.

In the same month, the administration of President Joe Biden began examining a similar issue.