Hello, tech enthusiasts! Today we dive into some significant shifts in AI policy as Anthropic has put forth its recommendations to the White House.

A day after quietly removing earlier commitments related to AI policy from its website, Anthropic has laid out suggestions for a national AI policy. They believe these steps will help America seize the economic benefits AI can bring.

One of the vital recommendations is to maintain the AI Safety Institute established under the previous administration. Anthropic also wants the National Institute of Standards and Technology (NIST) to conduct national security evaluations for advanced AI models.

In addition, the company suggests creating a specialized government team to assess potential security weaknesses in AI systems. Another pressing concern includes tightening the export controls on AI chips, particularly targeting the sale of Nvidia’s H20 chips to China, citing national security interests.

To boost the infrastructure supporting AI, Anthropic proposes an ambitious goal for the U.S.: to construct 50 additional gigawatts of power dedicated to AI by 2027.

Many of these policy suggestions echo the AI Executive Order initiated by the previous administration, which was rolled back in January. Critics during that time pointed to the order’s requirements as burdensome.