Washington charter for self-regulation: AI giants write their rules in the White House

The White House gathered leaders of major U.S. technology companies in a closed meeting chaired by Donald Trump, including the heads of Google, Anthropic, Meta, SpaceX, Nvidia and the chief of OpenAI, resulting in a self-regulation document described as ethically binding and exempting developers from any strict legislative framework imposed from outside the sector.

The agreement, signed by Trump alongside Sundar Pichai, Dario Amodei, Mark Zuckerberg, Greg Brockman, Elon Musk and Jensen Huang, states that the companies themselves must assume responsibility for ensuring the safety of their technical models. The companies pledged to implement internal protection measures to detect problems and fix them quickly, to cooperate with independent auditors to assess system efficiency and verify that they do not reach external technical systems or get breached unintentionally, while Trump announced his intention to create a government council to oversee the safety of these tools without specifying its composition.

The document leaves laboratories free to define safety standards and strips oversight of any legal consequences when failures occur.

The charter immediately drew criticism from researchers and academics in AI governance, with legal experts from the University of Sydney and the University of Melbourne deeming the agreement a distraction that sets only minimal safeguards, especially since it does not obligate companies to any explicit legal liability in the event of serious technical incidents. The accord came after OpenAI announced the cancellation of its latest model, GPT-6.1 Astra, over safety concerns, following an intensive audit after its models were breached into Australian government platforms without official authorization.

The meeting’s outcomes were not limited to the safety charter, as Trump reiterated his support for rapid expansion of data-center construction, viewing them as the backbone of advanced computing, despite growing public opposition in the United States over their massive electricity consumption. A Marist Institute poll showed that 65 percent of registered voters oppose building data centers in their areas, a stance the U.S. president met with a promise that companies would make substantial contributions to local communities, while Nvidia chief Jensen Huang described those centers as super-intelligence factories that help remanufacture America and create jobs.

In a parallel symbolic and administrative move, Trump signed an executive order requiring U.S. federal agencies and departments to replace the term “superintelligence” with “artificial intelligence” in official correspondence, reports and government websites, based on the view that current capabilities exceed the traditional definition of artificial intelligence.

This shift in the U.S. regulatory stance places policymakers and tech executives in the Gulf, Egypt and the region before a different operational reality. When Washington abandons binding regulatory constraints and leaves safety standards to the providers’ self-ethical commitment, information security and compliance officers in regional institutions can no longer rely solely on safety data issued by U.S. companies. Using these models in banks or government infrastructure now requires building internal independent testing and audit frameworks, and mandating sovereign-audit clauses in cloud-service contracts to protect data from unauthorized access.

The growing strain on U.S. power grids and the local resistance to data centers there also accelerate regional AI projects’ need to localize computing capacity and build domestic data centers that leverage abundant energy and ensure full operational sovereignty away from the domestic policy fluctuations of overseas developers.