Six AI companies sign a voluntary accord on controls and audits at the White House

US President Donald Trump and the heads of six technology companies signed the "White House Accord on Super Intelligence" at the White House on September 29, a voluntary commitment on the safety of the most advanced artificial intelligence models. The signatories are Sundar Pichai (Google), Dario Amodei (Anthropic), Mark Zuckerberg (Meta), Greg Brockman (OpenAI), Elon Musk (xAI) and Jensen Huang (Nvidia), following a meeting with members of the Cabinet and the Speaker of the House of Representatives, Mike Johnson.

The text starts from the premise that each company is responsible for developing its technology safely and proposes four layers of control. The first consists of internal controls to monitor the capabilities of the models and ensure they behave as intended during training and deployment, in areas such as cybersecurity, biosecurity and chemical threats, and to prevent them from accessing technical systems in unintended ways. The second is an internal team that checks these controls are working and fixes any problems. The third is an independent external auditor or evaluator. The fourth is an independent committee of the board of directors to oversee the process.

The companies also agree to meet regularly to set standards and best practices. The document sets no deadlines and notes that, over time, it may make sense to turn these measures into laws.

Johnson described it as a statement of principles with voluntary commitments and pointed out that Congress has not reached a consensus on regulating this technology, despite the work of bipartisan groups in both chambers during 2024.

On the same day, Trump signed an executive order requiring federal agencies to use, to the extent permitted by law, "Super Intelligence" and "SI" instead of "artificial intelligence" and "AI" in their official documents. According to the White House, the new name better reflects the current capabilities of this technology, which in its view is not limited to imitating or automating human intelligence. For now, the existing legal definition remains in place, and the presidential science and technology adviser has 60 days to propose a new one.

Key points

• Google, Anthropic, Meta, OpenAI, xAI and Nvidia sign a voluntary AI safety accord with Donald Trump.
• The document proposes four layers of control for the most advanced models.
• It includes internal controls against cybersecurity, biosecurity and chemical threat risks.
• It provides for an internal verification team, an independent external auditor and a board committee.
• The companies will meet regularly to set standards and best practices.
• The accord sets no deadlines and considers turning these measures into laws in the future.
• Mike Johnson describes it as a statement of principles given the lack of consensus in Congress.
• An executive order requires the Administration to replace the term AI with "Super Intelligence".
• The existing legal definition remains, with 60 days to propose a new one.