Independent, Expert, Accountable: What Third Party AI Oversight Needs to Actually Work

Articles

•

Lawmakers, labs, and the White House are converging on independent evaluation of AI. Whether it works depends on these criteria.

Independent verification organizations (IVOs), outside experts designated by government to test AI systems against public standards, are becoming central to AI governance. Fathom CEO Andrew Freedman lays out what they need to work. First, they have to be independent, with clear rules on who they can employ, who pays them, whether they can hold equity, and what access they get. Second, they need the right expertise, verified through credentials and matched to specific risk areas like biological risk, child safety, and loss of control. Third, they have to be accountable, with government designating them under public criteria, revoking licenses when they miss serious problems, and building a competitive market that rewards rigor. The test that matters most is whether an evaluator answers to the public or to the company it is assessing.

Independent.
Nonpartisan.
Nonprofit.

Fathom is a 501(c)(3) organization funded by philanthropists. We do not take donations from corporations, including frontier labs and the FAANG companies, or foreign entities associated with countries of concern.

Independent.
Nonpartisan.
Nonprofit.

Fathom is a 501(c)(3) organization funded by philanthropists. We do not take donations from corporations, including frontier labs and the FAANG companies, or foreign entities associated with countries of concern.

Independent.
Nonpartisan.
Nonprofit.

Fathom is a 501(c)(3) organization funded by philanthropists. We do not take donations from corporations, including frontier labs and the FAANG companies, or foreign entities associated with countries of concern.