Trump Allows AI Giants to Write Their Own Rules

Donald Trump initiated the signing of the "White House Agreement on Superintelligence" by six leading technology companies: Meta, Google, OpenAI, Anthropic, Nvidia, and xAI. The document is advisory in nature and does not impose strict legal obligations on corporations.

A four-level control and audit system is expected to be implemented to assess model capabilities in cybersecurity, biological, and chemical threats. Companies are required to establish internal monitoring teams, engage independent external auditors, and appoint a board committee to review reports.

Regulation Through Self-Discipline

The document lacks clear definitions for criteria of "reliable" control and the frequency of auditor checks. Instead, companies commit to regular meetings to independently develop industry safety standards. The agreement's authors allow for these measures to be codified into law in the future.

New Terminology in Government Documents

In addition to agreements with business, Donald Trump signed an order changing official terminology. Federal agencies are mandated to use the term "superintelligence" (SI) instead of the widely accepted "artificial intelligence" (AI) in all documents and on government websites. A precise definition of superintelligence must be developed within 60 days.