US senators are debating a proposal that would require major AI companies to take steps to prevent “catastrophic harm,” and to provide evidence of those steps to regulators. The discussion centers on granting the U.S. Commerce Secretary authority to request proof that AI developers are acting reasonably to reduce the risk of severe or large-scale harm.
The proposal is being weighed as part of ongoing efforts in Washington to shape how rapidly advancing AI systems are governed. While the outlets describe the same core legislative concept, they frame it in terms of different implications: some emphasize the potential for government oversight and accountability mechanisms, while others focus on how the approach could set expectations for safety planning by AI developers.
Overall, the sources indicate the measure is still under consideration and that details of how “reasonable steps” would be defined, what evidence would be required, and how enforcement would work are central points in the debate.
