Trump Signs Voluntary AI Safety Accord at White House Summit

Public concern over AI risks has intensified after reports of models bypassing controls and accessing systems without authorization. President Trump addressed that unease on September 29, 2026, by hosting executives from Alphabet, Meta, Nvidia, Anthropic, OpenAI and other firms at the White House for the signing of a voluntary accord.

The two-page White House Accord on Super Intelligence requires each company to set up internal monitoring of its models, maintain a team to verify those controls, engage outside auditors, and assign an independent board committee to oversee the process. Trump described the document as morally binding and likened it to a constitution. He also signed an executive order directing federal agencies to replace references to artificial intelligence with super intelligence in official communications where legally permitted.

Dario Amodei of Anthropic told reporters that the technology offers medical benefits and other gains but carries very real risks, adding that companies must work with the administration to win safely. Mark Zuckerberg of Meta called the accord a starting point that could build public confidence through internal reviews and external audits. Sundar Pichai of Alphabet termed the gathering historic.

A national survey by Verasight in September found 63 percent of respondents favored slowing AI development. Senator Mark Warner introduced legislation requiring rigorous testing and security reporting before powerful models are released. The accord contains no legal penalties for noncompliance, and several executives who previously called for regulation participated in the voluntary framework.

Trump posted the signed document on Truth Social. One copy released by the White House contained a misspelling of United States beneath his signature. Elon Musk briefly referred to the technology as AI before correcting himself to SI during remarks on potential abundance from the systems.