The White House Super Intelligence Accord was unveiled on September 29, 2026 (US Eastern Time). President Donald Trump published a one-page document on Truth Social titled "White House Accord on Super Intelligence: Joint Commitment on Frontier Responsibilities," calling it "close to constitutional" and a "moral constraint" on frontier AI companies, according to Reuters. Though only a page long, the document puts frontier AI companies' safety responsibilities for the super-intelligence era into words under the White House's name.
The signatories are Trump himself, plus six company leaders: Google CEO Sundar Pichai, Anthropic CEO Dario Amodei, Meta CEO Mark Zuckerberg, OpenAI President Greg Brockman, Elon Musk of X, and NVIDIA CEO Jensen Huang. Notably, Trump previously dismissed AI risk warnings as a "hoax," but has now pivoted to championing "strong self-regulation" — a shift driven by recent public concern over multiple AI agent attacks on federal databases.
What the accord covers: four layers of safeguards
The accord asks signatories to build four layers of safeguards: first, internally monitor models' cybersecurity, biosecurity, and chemical-threat capabilities during training and deployment, ensuring models do not "intrude into or access technical systems in unintended ways"; second, internal teams must keep control, monitoring, and detection measures running effectively; third, bring in independent external auditors to assess whether the control systems work as intended; fourth, establish an independent committee under the board of directors to receive reports from internal and external audits and frontline teams. The companies also commit to meeting regularly to develop safety standards and best practices.
The weight of a "moral constraint": no penalties, no deadlines
The accord states that "these steps may in the future be codified into law or regulation," but the full text contains no deadlines, no penalties, and no third-party regulatory role. In plain terms, this is a self-regulation pledge: companies face reputational pressure but no legal consequences, while the government gets a statement but no enforcement tools. A "moral constraint," translated into plain language, means it relies entirely on voluntary compliance.
Who is affected, and where the controversy lies
In the short term, the signatories' governance structures will be rewritten: an independent board committee and independent external audits are not cheap gestures, and once implemented they are hard to roll back. In the longer term, it could become a de facto industry standard, putting non-signatories at a disadvantage in public opinion and procurement. The controversy has two layers: first, the government's absence — the White House sets the stage but does not enforce, leaving the regulatory vacuum intact; second, credibility — companies grading themselves, with the independence of external audits depending on who pays and who defines the scope. Most importantly, it cannot restrain the model capability race itself: no matter how detailed the safety standards, they will not stop the next generation of models from shipping faster.
