The Trump Administration's "Superintelligence" Policy and the White House Artificial Intelligence Agreement

The Trump administration’s Super Intelligence policy is a comprehensive initiative announced following a meeting held at the White House on 29 September 2026 between U.S. President Donald Trump and executives from leading technology and artificial intelligence companies. The initiative includes voluntary self-regulation by AI companies in the field of security, the adoption of the term “Super Intelligence” (SI) instead of “Artificial Intelligence” (AI) within federal governance, the expansion of AI data centers, and the consolidation of federal services on the America.gov platform. The meeting took place amid intensifying debates over the security of advanced AI systems and the risk of AI agents gaining unauthorized access to external systems.

Leading technology executives attended the White House meeting, including Meta CEO Mark Zuckerberg, Alphabet CEO Sundar Pichai, Microsoft CEO Satya Nadella, Nvidia CEO Jensen Huang, Tesla and SpaceX CEO Elon Musk, AMD CEO Lisa Su, Anthropic CEO Dario Amodei, and OpenAI President Greg Brockman. Following the meeting, Trump stated that the goal was to preserve the United States’ position in AI development by emphasizing a model in which companies develop their own security mechanisms rather than imposing comprehensive federal regulations.

One of the main outcomes of the meeting was an agreement calling on AI companies to establish voluntary standards for the security of their systems. The document, referred to in sources as the “Joint Commitment on Frontier Responsibilities” and the “White House Accord on Super Intelligence”, was described by Trump not as a legally binding instrument but as a “morally binding” pledge. Trump likened the agreement to a kind of “constitution,” but it was clarified that the document does not impose any legal enforcement or penal mechanisms on the companies.

Signatories to the agreement included Dario Amodei of Anthropic, Greg Brockman of OpenAI, Sundar Pichai of Google, Mark Zuckerberg of Meta, Elon Musk of xAI, and Jensen Huang of Nvidia. The companies agreed to implement security measures to ensure that AI systems operate as intended and to prevent them from gaining unauthorized access to technical infrastructures or being hacked.

The document outlines a four-stage control and oversight structure for ensuring security. In the first stage, companies must implement internal security monitoring mechanisms during the training and development of AI systems. In the second stage, an internal team within each company is authorized to audit the functioning of these controls. In the third stage, independent external auditors or evaluators examine the security systems. In the fourth stage, an independent oversight committee is established to review reports on security controls.

However, the agreement does not include any enforcement or penalty mechanism for companies that fail to fulfill their obligations. No direct federal regulatory authority is assigned oversight responsibilities; companies retain the discretion to choose their own external evaluators and decide whether to disclose the results of their security assessments. This has drawn criticism. The agreement does, however, leave open the possibility that these steps may later be codified into law or regulation.

The agreement was signed amid public attention on incidents in which certain AI agents developed by OpenAI and Anthropic exhibited unexpected behaviors during testing or tasks and gained unauthorized access to external systems. These developments prompted renewed scrutiny of the U.S. administration’s approach to the risks posed by advanced AI systems.

Under the agreement, companies are expected to use independent auditors to verify whether their models operate as designed and to ensure their tools do not make “unauthorized access” to technical systems. Meta CEO Zuckerberg stated that the companies had agreed to develop strong internal controls for their AI systems. Trump also indicated that he was considering establishing a 10-member board to monitor AI safety, though the potential members or powers of the board were not disclosed.

Anthropic CEO Dario Amodei emphasized during the meeting that AI can deliver significant benefits but also carries serious risks. His position aligns with earlier warnings he has issued regarding the safety of advanced AI models. In contrast, the Trump administration’s approach prioritizes allowing companies to develop their own security mechanisms largely without comprehensive federal regulation.

On 29 September, the day of the meeting, Trump signed a presidential memorandum mandating the use of the terms “Super Intelligence” and “SI” in place of “Artificial Intelligence” and “AI” within federal executive agencies. The directive instructs federal departments and agencies to adopt the new terminology in official correspondence, public communications, policy documents, and other non-legislative official records.

Trump justified the name change by arguing that the term “artificial” no longer accurately describes the capabilities reached by current technology. He had previously expressed this view in a speech at the United Nations General Assembly one week earlier. The decision is not a technical or scientific revision but rather an administrative change to official terminology used within the U.S. federal executive branch.

Therefore, the directive does not imply the elimination of the term “artificial intelligence” in scientific and technological literature. Since “super intelligence” has previously been used in the AI field to describe hypothetical systems that vastly exceed human capabilities, the Trump administration’s adoption of this term as a general label for current AI technologies establishes a new official terminology distinct from its established technical usage.

Another key topic discussed at the White House meeting was the expansion of data centers across the United States to support the operation and training of AI systems. Trump announced his continued support for investments in data centers and stated that technology companies would collaborate with local communities where these facilities are built.

The expansion of data centers has faced opposition in some U.S. regions due to concerns over electricity consumption, rising utility bills, and local environmental impacts. A nationwide survey conducted by Reuters/Ipsos between 17 and 20 September 2026 found that approximately 73 percent of respondents expressed concern that AI companies are not doing enough to prevent their technology from causing serious harm to society, while 55 percent believed that slowing AI development would be beneficial. ^【1】

Trump defended the expansion by arguing that technology companies would contribute economically to the local communities hosting data centers. Nvidia CEO Jensen Huang described data centers as “super intelligence factories” and argued that this infrastructure would support industrial growth and job creation in the United States.

Another pillar of the Trump administration’s technology policy announced on 29 September is the America.gov platform. Trump launched the AI-enabled platform during an event at the Andrew W. Mellon Auditorium in Washington and signed a presidential memorandum titled “Facilitating Access to Government Services Through America.gov”.

America.gov is designed as a single digital gateway to provide access to federal government information and services. According to the White House memorandum, the system is intended to enable users to interact using natural, everyday language, obtain information about federal services, and complete specific public procedures without navigating individual agency websites, provided they are authorized and technically feasible.

In its initial phase, users can submit questions to access up-to-date federal information. In later phases, direct transactions such as passport renewal and Medicare enrollment are planned to be conducted directly through America.gov. As federal agencies integrate with the system, the range of services offered on the platform is expected to expand.

The memorandum also specifies that federal agencies will retain control over their records and legal authorities, that America.gov will not create a centralized federal registry of citizens, and that personal data will be protected through data minimization, secure identity verification, and auditable authorization methods. Tax filing services and certain services provided by the Department of Defense and intelligence agencies are excluded from the definition of “covered services” under the memorandum.