Early access for US government from AI giants

Under a new agreement between Google, Microsoft, and xAI and the US government, the most advanced AI models from these companies that have not yet been released to the public will be subject to national security assessments. The US AI Safety Institute (AISI), part of the Department of Commerce, announced that it has launched this initiative with the goal of deepening cooperation with the industry.

In a statement released by the AISI, it was noted that with the agreements signed with Google's DeepMind unit, Microsoft, and xAI, the most advanced AI technologies will be comprehensively examined before their public release. The institute has completed more than 40 evaluation processes, including many models that have not been published to date.

This collaboration allows companies to submit their models to the AISI by reducing or completely removing safety measures, thereby enabling an objective analysis of the impact of these models on national security. The aim is to take precautions against potential dangers, especially in AI software that has not yet been launched on the market.

AISI Director Chris Fall stated in his remarks on the subject, "Independent and rigorous measurement science is essential to understand the most advanced AI and its impact on national security. These expanded industry collaborations help us scale our work in the public interest at a critical moment."

The new agreements aim to ensure that AI models undergo a serious preliminary assessment regarding national and societal security before they are used on a large scale. By deepening this type of industry-government cooperation, the US aims to reduce the risks of technological advancement and protect its citizens from potential threats.

News Source: 12punto