White House AI accord leaves the definition of an independent auditor open

Google, OpenAI, Anthropic, Meta, xAI and Nvidia signed the White House Accord on Super Intelligence, committing to four layers of frontier-model oversight including an independent external assessor. The accord does not define who qualifies as independent, what standards they use or how much access they get.

• The accord sets four oversight layers: internal controls, internal verification, an external assessor and a board committee
• Commitments are voluntary for now but could later be written into law
• Each company picks its own assessor, with no shared independence criteria
• Connecticut will launch a pilot in July 2027 approving up to five verification organizations