The White House Accord on Super Intelligence, released on September 29, 2026, establishes a voluntary framework for AI governance. Signatories include executives from Anthropic, OpenAI, Google, Meta, xAI, and Nvidia, alongside President Trump. The agreement mandates internal risk controls, dedicated monitoring teams, independent external audits, and an independent board committee.
Schellman identifies a critical flaw in the accord: the lack of a formal definition for independent auditors. The framework does not specify qualification criteria, independence standards, or conflict-of-interest management. Consequently, each company determines its own compliance methods, leading to significant inconsistencies in how these commitments are implemented.
Data from Schellman’s July 2026 State of AI Governance Report highlights a gap in readiness. While 74% of professionals believe their organizations could pass an audit, only 27% possess mature governance programs. Furthermore, 90% of organizations have dedicated funding for governance, yet only 57% have formal policies and 44% have incident response procedures.
Deployment of AI agents is outpacing governance development, with 86% of companies testing agents and 46% utilizing them in production. The accord lacks government oversight mechanisms, public disclosure requirements for audit results, and penalties for non-compliance. This places the burden of verification on investors and regulators to assess the validity of self-reported readiness.
