The AI Fraud Loophole Washington Has Chosen Not to Close

Summary

This opinion essay argues that US corporate-registration rules leave an exploitable opening for current AI systems: in most states, a company can be formed without verifying the identity of its owners or controllers. Because AI can produce paperwork at negligible cost, it could help fraudsters create shell-company networks, while existing structures also enable practices such as “phoenixing” and trucking “chameleon carriers” that shed debts or safety records. The author says Washington widened the gap when FinCEN permanently exempted US-formed companies from reporting their beneficial owners and said it would delete previously filed, unverified records. The proposed remedy has two parts: require at least one controlling person to verify their identity in person when a company is formed, or within a year for existing companies; and make that person personally responsible for fraud judgments or penalties if the company disappears to avoid them. The essay argues that in-person verification would make stolen identities and nominal front owners harder to use, while criminal penalties for knowingly acting as a front would address the remaining workaround. It cites UK director verification, Australia’s director-ID system, federal court precedent on ownership disclosure, and existing US payroll-tax liability as evidence that the approach is feasible. The author also extends the identity-and-liability principle to payment systems, cryptocurrency spending, and access to powerful AI for legitimate cyber defenders, citing Anthropic’s defender verification program as an example. The piece is an argument rather than an empirical evaluation, and it acknowledges that controls may need to change if AI finds routes to power that never pass through an accountable person.