Leaders from six AI companies signed a White House accord on Tuesday committing to layers of internal controls, outside assessments and board oversight for frontier models. The original “Joint Commitment on Frontier Responsibilities” document lists Sundar Pichai for Google, Dario Amodei for Anthropic, Mark Zuckerberg for Meta, Greg Brockman for OpenAI, Elon Musk for xAI and Jensen Huang for Nvidia.
The commitments cover controls during both training and deployment. The document names cybersecurity, biosecurity and chemical threats, as well as preventing models from hacking or accessing technical systems in unintended ways.
Its four layers start with those internal controls and an internal team responsible for checking that monitoring and detection work and that problems are fixed. Companies also commit to working with an independent external auditor or evaluator. An independent board committee would oversee the controls teams and audit reports, including remediation of identified issues.
The participating companies plan to meet regularly to establish standards and best practices. Pichai confirmed signing the accord and pointed to Google’s testing, evaluations, red teaming and safeguards.
The accord describes commitments rather than newly enacted requirements. Its text says the steps may eventually make sense to codify into laws or regulations, while committing the companies to them regardless. The signing establishes those promises; it does not demonstrate that the promised controls or independent assessments have been completed.
